PROCEDURE FOR THE DETERMINATION OF TUNA BIOMASS IN A WATER AREA AND CORRESPONDING SYSTEM (Machine-tr

专利摘要:
Procedure for the determination of tuna biomass in a water zone and corresponding system. Submerged under said water zone (1) are measuring means (2) with an acoustic sensor and an image capture means, at known positions and directed in an upward direction. With the following stages: said acoustic sensor performs acoustic measurements generating an echogram; said image capture means capture a sequence of images; the number of tunas that have crossed said water zone (1) is determined using said echogram; the size and weight of the tunas is determined using said sequence of images; and a biomass value of tuna is determined from the number of tunas and the weight of the tunas. (Machine-translation by Google Translate, not legally binding)
公开号:ES2665583A1
申请号:ES201631379
申请日:2016-10-26
公开日:2018-04-26
发明作者:Gabriela ANDREU GARCÍA；Vicente ATIENZA VANACLOIG；Víctor ESPINOSA ROSELLÓ；Vicente Domingo ESTRUCH FUSTER；Pau MUÑOZ BENAVENT；Patricia ORDÓÑEZ CEBRIÁN；Isabel PÉREZ ARJONA；Vicente PUIG PONS；Ester SOLIVERES GONZÁLEZ；José Miguel VALIENTE GONZÁLEZ；Eladio SANTAELLA ÁLVAREZ；Fernando DE LA GÁNDARA GARCÍA
申请人:Instituto Espanol De Oceanografia (ieo)；Balfego & Balfego S L；Balfego & Balfego SL；Inst Espanol De Oceanografia Ieo；Universidad Politecnica de Valencia；
IPC主号:

专利说明:

Field of the Invention
The invention is in the field of sustainable fish farming activities.
More specifically, the invention relates to a process for the determination of tuna biomass in a water zone. The invention also relates to the corresponding system. In the context of this invention, the concept of biomass also extends to measures such as the number of individuals and their size.
15 State of the art
During the last years a great effort has been made by the authorities and especially by fishermen and producers to favor the recovery of the stock 20 of bluefin tuna. This has involved taking measures in terms of the timing of the calendar, minimum size and fishing quotas. One of the existing systems is based on the capture of live adult specimens once they have spawned, and their subsequent fattening in marine cages, with controlled feeding, so that the specimens can be sacrificed according to market demand, thus equalizing the 25 number of catches and reducing their impact in critical phases of tuna growth. In this way, tunas are located and captured by surrounding them with fencing nets and then transferred to transport cages and from there to fattening farms. For the sake of conservation, the governments of each zone, for example within the ICCAl (International Commission for the
30 Conservation of Atlantic Tuna) regulate the maximum number of specimens and biomass that can be captured, so it is essential to count and estimate the size that allows the control of these quotas.
The current commercial systems are based on a count of number of copies and
its size based on an image analysis, either automatic, with assistance
human limiting the size of the silhouette of tunas, or even completely manual.
A widely known example is the VICASS® system. From the size it
5 performs a weight calculation using an algorithm established from the
fishing regulations of each geographical area.
However, these types of systems have a number of drawbacks, such as
the time required to obtain the results, their accuracy, and the
10 variability in results that depends strongly on the person attending the
system and how optimistic or pessimistic it is by manually limiting the size of the
if lueta of the copies.
Description of the invention
fifteen
The invention aims to provide a method for the determination
of tuna biomass in a water zone of the type indicated at the beginning, which allows
solve the problems outlined above. Another object of the invention is
Provide a corresponding system.
twenty
This purpose is achieved through a procedure for determining
tuna biomass in a water zone of the type indicated at the beginning, characterized
by means of measuring means which are arranged under said water zone
they comprise an acoustic sensor and image capture means, being
25 said measuring means in known positions and directed in the direction
ascending, the procedure comprising the following steps:
during a measurement interval in which some tunas cross this area of
Water:
said acoustic sensor performs acoustic measurements by emitting
JO acoustic signals and receiving echoes, thus generating an echogram; Y
said image capture means perform image captures,
generating a sequence of images containing images of said
tunas;
determine a value of the number of tunas that have crossed this area ofwater during said measurement interval using said echogram;determining a tuna weight value using said sequence of images;determine a biomass value of tunas that have crossed this area of
5 water during said measuring range from said value of the number of tunas and said weight value of tunas.
In the context of the invention said water zone is the measuring zone through which tunas go through, it can have different forms depending on the characteristics of the application and the measuring means. Thus, since the measuring means are focused in an upward direction, although not necessarily completely vertical, given the normal way of swimming of the tunas, ventral measurements are obtained in which the background of the images is the maximum clear due to the sunlight and tunas result in shadows, which increases contrast and facilitates detection and measurement. In addition, it has also been found experimentally that this perspective of the tuna body is quite stable despite its movements and turns when swimming. At the same time, it has been found experimentally that carrying out ventral acoustic measurements makes it easy to identify specimens for two reasons: the first is that the maximum echo is achieved from the swim bladder and from the ventral zone, the sound must pass through less complex structures to said bladder, reaching more uniform. In contrast, the dorsal measures tend to cause a series of echoes of more complex structures such as the skeleton of the fish, the skull, etc. together with internal rebounds that can make it difficult to obtain a suitable model for identification. The second reason is that to identify individuals, it is convenient to place the probe at a sufficient distance to avoid the effects of the near field, so that an echo with a wavefront as spherical as possible is received. Thus, depending on the application, the fish are very close to the surface, which makes it impossible to avoid the near field with dorsal acoustic measurements. Another added benefit of ventral measures is that, in this case, waves 30 bounce in the area with more fat accumulation, which also allows estimating the so-called tuna condition factor, that is, the amount of fat they present . In the context of this document, echography is referred to as a
representation of acoustic echo measurements as a function of time, not just for an instant.
The combination of acoustic and optical measurements has the effect of obtaining the best of
5 both, thus the acoustic measurements allow to quickly identify the number of specimens, even in turbidity conditions, while the optical measurements, with the maximum available contrast, allow to see the sizes of the tunas and infer their weight. The combination of both types of measures may even become dependent, so that correlations of specimens are identified between the
10 echogram and sequence of images.
On the basis of the invention defined in the main claim, preferred embodiments are provided whose characteristics are set out in the dependent claims.
Preferably, determining said tuna weight value comprises: selecting tuna from said sequence of images; determine a size value calculation for each of said selected tunas;
20 determining an individual weight value calculation for each of said tunas selected from said size value; and determining said tuna weight value as the average value of said individual weight value for each of said selected tunas; wherein said biomass value is determined as the multiplication of said value
25 of tuna numbers for said tuna weight value. For the sake of simplicity, in the context of this document and, unless otherwise indicated, size referred to the dimensions, as a whole or a combination thereof, will be discussed. Local regulations usually indicate conversion parameters between size and estimated weight, so this is a calculation
Approximately 30 but accepted in the art. Thus, tunas are selected from the sequence of images and their size is measured for said tunas, obtaining a weight. Depending on the formula for the final calculation of biomass, all tunas or only those considered most representative can be measured, even avoiding those whose silhouette is confusing, in an overlapping situation, etc.
Preferably, said image capture means comprise two sensors
5 optics located at a relative position between them, such that said sequence of images comprises a stereoscopic sequence comprising a first sub-sequence of images and a second sub-sequence of images, both sub-sequences being synchronized; wherein said procedure comprises the additional steps of:
10 selecting tuna present simultaneously in said first sub-sequence of images and in said second sub-sequence of images; determining a distance value calculation for said selected tunas, using the values of said relative position between said first optical sensor and said second optical sensor; Y
15 modify said size value for each of the selected tunas
using said distance value; In this way, a sequence of stereoscopic images is obtained, of which the relative position between the optical sensors is known, for example video cameras. In each frame there are two images from different angles, which allows
20 also infer the distance at which the fish is. This is especially advantageous since tunas can move to multiple levels and using a medium distance can have a negative impact on the quality of the measure. By using two cameras, distance measurement is improved and the optical image can even be correlated with the echogram. For the sake of simplicity, the details are not detailed here.
25 calibration procedures for stereoscopic imaging systems since they are widely known in the art and must be performed prior to measurements.
Preferably, select tuna present simultaneously in said
First sub-sequence of images and in said second sub-sequence of images comprises selecting two characteristic points of each tuna selected in one of said first sub-sequence and said second sub-sequence, said two characteristic points being preferably the tip of the head and the tip of the tail, and projecting said two points on the other between said first sub-sequence and said second sub-sequence, through epipolar lines; so that those tunas with said two characteristic points are selected in one of said first sub-sequence and said second sub-sequence that correspond to
5 other two characteristic points located in said epipolar lines in the other between said first sub-sequence and said second sub-sequence. The use of epipolar geometry is known to extract three-dimensional position information from stereoscopic images. This selects those tunas that appear in the two sub-sequences and whose points are aligned
10 according to said polar lines, which is a necessary condition for it to be the same element. In this preferred form, the choice of two points takes into account the swimming direction and the position of use of the image capture means, so that when using the head and tail points, the maximum distance for the fish is worked , which improves accuracy.
Preferably, determining a distance value calculation and modifying said size value for said selected tunas comprises the steps of:
correct distortion effects of the lenses in said sequence of images;
20 obtaining the distance value by triangulation of each pair of points of said sequence of images, wherein said pair of points comprises a point in one of said first sub-sequence and said second sub-sequence, and a corresponding point in the other between said first sub-sequence and said second sub-sequence; Y
25 modifying said size value for each selected tuna based on said distance value of said pairs of points corresponding to said selected tuna. Thus, the correction of optical defects of the lenses allows the use of wide-angle lenses or even fisheye calls, which typically have a strong
30 distortion, especially at its ends and usually barrel-shaped. Using direct images would imply inaccuracies in determining size. On the other hand, a corrected form of tuna is obtained as a function of distance, which affects the accuracy of the size determination.
Preferably, said sequence of images comprises a plurality of
frames and in which to select some tunas from said sequence of images
It comprises the following stages:
5 segment said frames to obtain first candidate objects; filter said first candidate objects by their geometric characteristics, obtaining a few second objects that are a subset of said first objects;
Thus, said candidate objects are segmented by methods known in the art, by
10 example, by using the so-called blobs. Given the characteristics of the images, the light intensity value may not be constant, which could be difficult to determine the silhouette of the tuna, for this reason the use of crazy contrast is a preferred solution, since in any case, the Bottom is lighter than fish. This strategy allows an effective cut. They are then removed
15 those objects that, due to their geometric characteristics such as dimensions or shape, are too far away from what a tuna might be, for example, echoes related to the structures of the nets, ships, etc.
Preferably, select tuna from said sequence of images
20 comprises the additional step of: adjusting a tuna image model to said second objects, so that if for an object the adjustment error exceeds an error threshold value, said object is discarded, whereas otherwise it is considered that said object is a selected tuna and an adjusted model is obtained;
25 and wherein determining a size value calculation for each of said selected tunas comprises obtaining said size value from said adjusted model of said selected tuna. Thus, it starts from a model that represents the shape of a tuna and tries to adjust that model on the selected candidate objects. If the object does not fit the model of
30 tuna is discarded, that is, an adjustment error is calculated based on the type of model and if that error exceeds a threshold, the object is discarded. This threshold will then be dependent on the model itself, and it will be necessary for the expert to obtain an experimental threshold value than the best results. As a non-limiting example, if the model is a silhouette of points, the error may be the quadratic sum of the distances of those points to the silhouette of the candidate object. For objects that meet the criteria of the tuna model, the model is adjusted to the particular shape and that adjusted model is used to obtain the size, as a non-limiting example, a
5 tuna model that contemplates torsion can allow tuna size to be measured based on the torsion curve.
Preferably, said tuna image model comprises a segment model in which some points, called vertebrae, Vi, located on axis 10 of the tuna are used, preferably 20 points, numbered from vertebra O (va) corresponding to the tip of the head, to vertebra 19 (V19) that corresponds to the end of the tail. That is, each point O vertebra Vi of the model is located on the backbone of the model, and can have a uniform distribution or be variable, for example to better profile features such as fins or peduncle of the
15 fish
Preferably, an estimate of the tuna contour is made by relating each of said vertebrae, Vi, with an estimated contour point, ki, preferably by means of coefficients that relate the position of each of said vertebrae.
20 vertebrae Vi with the distance to vertebra Vi of each of said estimated points of the contour ki, preferably the following coefficients:
CU = [O, 0.7, 1.15, 1.35, 1.55, 1.65, 2.0, 2.5, 2.8, 3.15, 1.7, 1.55, 1.35, 1.1, 0.9, 0.7, 0.55, 0.45, 0.5, 0.75]
where the values are sequentially ordered from V to 9. Thus, a model is used in which the fish lengths also serve to determine the width. The coefficient values have been obtained in a way
25 experimental and correspond to a situation without flexion, in which the fish is straight.
Preferably, said model assigns a different contribution value of each vertebra given a global flexion, and, preferably following the following formulation:
¡= 19
0 * óO¡ "
I d9, = 9; dO¡ = I! J.O¡ = 16; 'r / v¡ E v;
A.O.O.
i = O; = 0
where the global flexion and the values of the coefficients or (Ji represent the
contribution of the flexion of each vertebra Vi to the global flexion e, preferably taking the following values: ~ 8 = [O, O, 0.64, 0.64, 0.48, 0.48, 0.0, 0.0, 0.0, 0.0, 0.48, 0.64, 0.8, 0.96, 1.12, 1.28, 1.44, 1.92, 2.4, 2.72]
5 where the values are sequentially ordered from Vo to V19. Thus it is possible to obtain a size measurement from the model even when the tuna is flexed. In addition to allowing better identification of fish in the images obtained by adapting to the shape of the tuna swimming according to the observations.
Preferably, said model also comprises translation and rotation parameters. With which the model can be rotated an angle and moved by the image, which allows to identify the fish in any position.
Preferably, determining said value of the number of tunas that have crossed said water zone during said measurement interval using said echogram comprises the steps of:
segment regions of said echogram by applying mathematical morphology criteria, considering that said echogram is an image;
20 obtain segmented regions; and determine said value of the number of tunas as the value of the number of said segmented regions.
In this way, the ultrasound is treated as if it were an image and image segmentation techniques are used to identify the regions likely to be a tuna.
Preferably, said mathematical morphology criteria comprise ultrasound trace models for tunas, so that said regions are segmented by morphological similarity with said trace models. Thus, a criterion is used in such a way that the echo of a tuna must have, which allows refining the result and even separating overlapping areas that would otherwise be complex. The morphological model of the ultrasound trace can be obtained based on periodic measurements of real tunas from which a location verification can be obtained, which
5 allows to improve this model and with it the segmentation of the echogram.
In the context of acoustic sounding measurements, it is known as target strenght, or TS for its acronym in English, a measure of the reflection coefficient of a sonar target. Generally its value is a negative number of decibels that varies with the
10 characteristics of the objective, in this case a tuna.
Preferably, control parameters are chosen that comprise one or more of the list consisting of: target strenght margins, TS; Y
15 margins of tuna size; where by segmenting regions of said echogram those regions corresponding to values outside said control parameters are discarded. In this way, those echoes that get out of these patterns are discarded, for example by having too much TS or too little, or by being too large or
20 small In this way you can eliminate the echoes coming from structures, boats, other fish, etc.
Preferably, it comprises the additional step of processing said echogram to compensate for an effect on said echogram caused by any of the parameters
25 included in the list consisting of: water temperature; water salinity; water pH;
Thus, measurements of said water parameters are made and the echogram is processed
30 to compensate for the effect that these physical parameters have on it, by influencing, for example, the speed of sound transmission, which would result in echoes that are in an apparent location closer to or farther from the real one.
Preferably, a support structure is provided that joins a first cage with a second cage, wherein said support structure has a first passage window that communicates with said first cage and a second passage window that communicates with said second cage, defining a passage area for tunas between said first window and said second window, a direction of passage that goes horizontally from said first cage to said second cage, thus defining a first direction that coincides with said direction of passage, and a second direction that it is horizontal and perpendicular to said first direction; wherein said structure presents a first profile looking in said first direction, and a second
10 profile looking in said second direction; and wherein said measuring range comprises a time in which a plurality of tunas is moved from said first cage to said second cage in said passage direction. Thus, the procedure can be used to calculate biomass in tuna transfer operations. In the context of the invention, one speaks interchangeably of
15 cage, purse seine or farm. The tunas pass from one cage to another through a structure configured for it and that serves as a physical support to hold the technical elements of the invention.
Preferably, said acoustic sensor has a detection beam in the form of
20 curtain with a wide side and a narrow side, said sensor arranged so that said wide side extends in said second direction presenting a second beam width such that it covers all said water zone, and said narrow side extends in said first direction presenting a first beam width less than 5 °, preferably less than 2 °, more preferably between 1 ° and 0.5 °.
25 Thus the acoustic sensor covers the entire passage area, so that it detects all the fish that pass through that area, but attempts are made to measure the fish only once while crossing the curtain. In the context of the technique it is considered that the beamwidth is defined by a power drop of -3dB
Preferably, said acoustic sensor emits a burst of pulses with a burst duration of less than 100 microseconds, more preferably less than 50 microseconds. The use of very short pulses together with a curtain beam that the fish go through allows to make "cuts" of acoustic response of the fish, with an echo that is clear even at high displacement and superposition speeds, as can happen with tunas, which improves spatial resolution and allows distinguishing cases of fish overlap. Alternatively or complementary, the use of frequency modulated pulses is preferred, together with correlation techniques
5 of the echo with the emitted pulse, what is known as the pulse compression technique. This further increases the signal-to-noise ratio, SNR, as well as the spatial resolution of the internal structure of the traces of each fish, thus improving the distinction between fish.
10 Preferably, said acoustic sensor emits pulses that are repeated with a frequency greater than or equal to 10 Hz, preferably between 30 and 40 Hz. Thus, for example, with a frequency of 30 Hz in which the cycle of bursts of detection pulses (pulse repetition rate, prr, in English), for fish with a length of 2m that travel at a maximum speed of 60 km / h each fish receives
15 minus two pulses, and therefore two cuts are generated on the echogram, which has been experimentally found to be a minimum value to guarantee detection as much as possible. The maximum limit is also imposed by the distance of the probe, for example, for a distance of 16m to the top of the fish passage area, the sound covers a total of 32m round trip, which at the speed typical of
20 propagation of sound in the water of 1500 mis gives 21.3 milliseconds. Thus, the maximum burst frequency (ppr) that avoids signal overlap between cycles would be around 46 Hz.
Preferably, said acoustic sensor comprises an echo sounder and a transducer of
25 wide beam, wherein said second beam width is between 40 ° and 80 °, preferably between 45 ° and 55 °, more preferably 49 °. Being these experimental parameters that are adapted to the experimentally more advisable measures of the area of passage of the tunas for the case of the Mediterranean waters and for the behavior of the bluefin tuna, as well as to the suitable depth to which they must
30 put on the measuring means. In the technique, areas of passage 10m deep by 10m wide are common. On the other hand, it is not convenient to place the image capture means too deep, since the turbidity of the water has an impact on the clarity of the images. Thus, greater depths are not convenient.
of 16m. By simple trigonometry, it is determined that in this case, a centered transducer should have a wide beam of 80 °, however the cost of a transducer with these characteristics can be very high. For this reason, a lower cost transducer is chosen, around 50 ° which, although it does not cover the entire surface, in practice only the lower corners are left out, where it is very unlikely that the given fish will pass Your behavior observed. In addition, the beamwidth is generally referred to as that angle delimited by a drop of -3dB, so, although attenuated, echoes of angular distances greater than those of the beamwidth continue to be received. Using the same criteria,
10 the expert will have no problem determining the appropriate values for other situations.
Preferably, said measuring means are positioned on the inner bottom of said support structure, centered with respect to said second direction, and in which said acoustic sensor is positioned directed in a vertical direction, so that said water zone is comprised in said passage area. Thus the location of the tunas is done by a narrow vertical curtain, so that an attempt is made to receive the most ventral echo possible. The centered position is especially advantageous if the structure is symmetrical with respect to said second direction with a passage area also
20 symmetric
Preferably, said measuring means are positioned centered with respect to said first direction. What is especially advantageous to move as far as possible the acoustic measurements inside these cages in which there may be tunas. This
25 affects less interference and also more appropriate images.
Preferably, said second profile of said support structure has a rhombus shape, with an upper vertex and a lower vertex, and in which said measuring means are positioned in said lower vertex and directed in a vertical direction. Thus, the shape of the structure causes it to move away from the measuring means in the area closest to them, minimizing interference with them. In turn, the narrow upper vertex of the structure minimizes the area of passage of tunas, so
It is easier for them to quickly move from one cage to another, without going around in the middle.
Preferably, said support structure has a second profile in the form of
5 pentagon with an upper vertex and a lower flat base, in which said measuring means are positioned in said flat base and in which said image capture means are positioned directed towards said first window or said second window. Thus avoiding interference from the flotation structures of the cages that meet precisely at the top vertex of the support structure, which results
10 in a lower contrast in the images due to the blocking of sunlight. Arranged in this way, the acoustic sensor measures vertically and the image capture means obtain these images with the background window through which sunlight enters without interference. In return, both elements are misaligned and the size measurement of tunas should take into account the angle at which they are
15 images taken.
Preferably, said second profile of said support structure has a
rectangular shape, with a flat bottom and top, and in which all
said measuring means are positioned in said lower part and directed in the direction
20 vertical. What gives a simple and shadowless arrangement, at the cost of a greater distance between cages. This arrangement is preferred in the case of specimens whose characteristics do not pose a problem for the passage from one cage to another through a relatively long corridor.
Preferably, said measuring means are arranged outside said support structure and directed in a vertical direction, positioned in said first cage so that said water zone is located in front of said first window, or in said second cage of such that said water zone is located in front of said second window. In this way, the image and the echogram can be vertical
30 and without shadows. This setting is preferred in cases where the cages are large enough so that the tunas that have already passed do not interfere with the count.
The invention also relates to a system for the determination of tuna biomass in a water zone, characterized in that it comprises measuring means comprising an acoustic sensor and image capture means, configured to be arranged submerged under said zone. of water, in positions 5 known and directed in an upward direction, in which during an interval of
measurement in which some tunas cross said water zone: said acoustic sensor is configured to perform acoustic measurements by emitting acoustic signals and receiving echoes, thus generating an echogram; Y
10 said image capture means are configured to perform image captures, generating a sequence of images containing images of said tunas;
said system also comprising analysis means configured to: determine a value of the number of tunas that have crossed said area of
15 water during said measurement interval using said echogram; determining a tuna weight value using said sequence of images; determining a biomass value of tunas that have crossed said water zone during said measurement interval from said value of the number of tunas and said tuna weight value.
20 The characteristics and technical effects of this system are equivalent to those of the procedure described above, so, for the sake of clarity, they will not be repeated here.
Preferably, determining said tuna weight value comprises: selecting tunas from said sequence of images; determine a size value calculation for each of said selected tunas; determine an individual weight value calculation for each of said
30 tunas selected from said size value; and determining said tuna weight value as the average value of said individual weight value for each of said selected tunas; wherein said biomass value is determined as the multiplication of said tuna number value by said tuna weight value.
Preferably, said image capture means comprise two sensors
5 optics located at a relative position between threads, such that said sequence of images comprises a stereoscopic sequence comprising a first sub-sequence of images and a second sub-sequence of images, both sub-sequences being synchronized; wherein said analysis means are configured to perform the additional steps of:
10 selecting tuna present simultaneously in said first sub-sequence of images and in said second sub-sequence of images; determining a distance value calculation for said selected tunas, using the values of said relative position between said first optical sensor and said second optical sensor; Y
15 modify said size value for each of the selected tunas using said distance value;
Preferably, selecting tuna present simultaneously in said first sub-sequence of images and in said second sub-sequence of images 20 comprises selecting two characteristic points of each tuna selected in one of said first sub-sequence and said second sub-sequence, said two characteristic points being preferably the tip of the head and the tip of the tail, and projecting said two points on the other between said first sub-sequence and said second sub-sequence, through epipolar lines; so that it
25 select those tunas with said two characteristic points in one of said first sub-sequence and second sub-sequence that correspond to two other characteristic points located in said epipolar lines in the other between said first sub-sequence and said second sub-sequence
Preferably, determining a distance value calculation and modifying said size value for said selected tunas comprises the steps of: correcting lens distortion effects in said image sequence;
obtaining the distance value by triangulation of each pair of points of said sequence of images, wherein said pair of points comprises a point in one of said first sub-sequence and said second sub-sequence, and a corresponding point in the another one between said first sub-sequence and said
5 second sub-sequence; and modifying said size value for each selected tuna based on said distance value of said pairs of points corresponding to said selected tuna.
10 Preferably, said image sequence comprises a plurality of frames and in which selecting tunas from said image sequence comprises the following steps:
segment said frames to obtain first candidate objects; filter said first candidate objects by their geometric characteristics, obtaining second objects that are a subset of said first objects;
Preferably, selecting tunas from said sequence of images comprises the additional step of:
20 adjust a tuna image model to said second objects, so that if for one object the adjustment error exceeds an error threshold value, said object is discarded, whereas otherwise it is considered that said object is a selected tuna and you get a tight model; and in which to determine a size value calculation for each of said tunas
25 selected comprises obtaining said size value from said adjusted model of said selected tuna.
Preferably, determining said value of the number of tunas that have crossed said water zone during said measurement interval using said echogram
30 includes the steps of: segmenting regions of said echogram by applying mathematical morphology criteria, considering that said echogram is an image; obtain segmented regions; and determine said value of the number of tunas as the value of the number of said segmented regions.
Preferably, said mathematical morphology criteria comprise about 5 echogram trace models for tunas, so that said regions are segmented by morphological similarity with said trace models.
Preferably, said analysis means are configured to allow the choice of control parameters comprising in one or more of the list that
10 consists of: target strenght margins, TS; and margins of tuna size;
wherein said analysis means are configured so that when segmenting regions of said echogram, discard those regions corresponding to values outside said control parameters.
Preferably, said analysis means are additionally configured to
process said echogram to compensate for an effect on said echogram caused by
any of the parameters included in the list consisting of:
20 water temperature; water salinity; Water pH:
Preferably, it further comprises a support structure configured to be
25 arranged in a use position in which it joins a first cage with a second cage, where said support structure has a first passage window that communicates with said first cage and a second passage window that communicates with said second cage, defining a passage area for tunas between said first window and said second window, a direction of passage that goes horizontally from
Said first cage to said second cage, thus defining a first direction that coincides with said direction of passage, and a second direction that is horizontal and perpendicular to said first direction; wherein said structure has a first profile looking in said first direction, and a second profile looking in said second direction; and in which said system is configured of
such that the measurement range comprises a time in which a plurality of
tunas move from said first cage to said second cage in said direction of
He passed.
Preferably, said acoustic sensor has a detection beam in the form of a curtain with a wide side and a narrow side, so that in said position of use said sensor is arranged so that said wide side extends in said second direction presenting a second beam width such that it covers all bliss
10 water zone, and said narrow side extends in said first direction having a first beam width less than 5 °, preferably less than 2 °, more preferably between 1 ° and 0.5 °.
Preferably, said acoustic sensor is configured to emit bursts of 15 pulses having a burst duration of less than 100 microseconds, more preferably less than 50 microseconds.
Preferably, said acoustic sensor is configured to emit pulses having a repetition frequency greater than or equal to 10 Hz, preferably
20 between 30 and 40 Hz.
Preferably, said acoustic sensor comprises an echo sounder and a wide beam transducer, wherein said second beam width is between 40 ° and 80 °, preferably between 45 ° and 55 °, more preferably 49 °.
Preferably, in said position of use said measuring means are positioned in the inner bottom of said support structure, centered with respect to said second direction, and in which said acoustic sensor is positioned directed in a vertical direction, so that said water zone is included in said zone
30 step.
Preferably, in said position of use said measuring means are positioned centered with respect to said first direction.
Preferably, said second rock of said support structure has a
rhombus shape, with an upper vertex and a lower vertex, and in which in said
position of use said measuring means are positioned at said vertex
5 lower, and directed in a vertical direction.
Preferably, said support structure has a second pentagon-shaped profile with an upper vertex and a lower flat base, in which in said position of use said measuring means are positioned in said base
10 flat and in which said image capture means are positioned directed towards said first window or said second window.
Preferably, said second rocker of said support structure has a rectangular shape, with a flat lower part and upper part, and in which in said position of use all said measuring means are positioned in said lower part and directed in vertical direction
Preferably, in said use position said measuring means are positioned outside said support structure and directed in the direction
20, positioned in said first cage so that said water zone is located in front of said first window, or in said second cage so that said water zone is located in front of said second window.
The invention also encompasses other detail features illustrated in the
A detailed description of an embodiment of the invention and in the accompanying figures.
Brief description of the drawings
The advantages and features of the invention can be seen from the following description in which, without limitation with respect to the scope of the main claim, preferred embodiments of the invention are set forth with reference to the figures.
Fig. 1 shows an example of a joint visualization of the sequence of images and the echogram for an embodiment of the invention.
5 Fig. 2 shows a graphic representation of how to correlate a tuna between two stereoscopic images through epipolar lines.
Fig. 3 shows a sequence of an example segmentation procedure for an image of the image sequence. 10 Fig. 4 shows an example tuna model.
Figs. 5a, 5b and 5c show a sequence of a process of adaptation to the tuna model of candidate objects. 15 Fig. 6 shows an example echogram.
Fig. 7 shows a segmented region of the echogram of Fig. 6.
Fig. 8 shows a schematic representation of an embodiment of the invention with a diamond-shaped structure.
Fig. 9 shows a schematic representation of an embodiment of the invention with a pentagonal structure and the image capture means 25 directed to the second window.
Fig. 10 shows a schematic representation of an embodiment of the invention with a pentagonal structure and the image capture means directed to the first window.
Fig. 11 shows a schematic representation of an embodiment of the invention with a rectangular structure.
Fig. 12 shows a schematic representation of an embodiment of the invention with the measuring means outside the support structure.
5 Fig. 13 shows a schematic representation of an embodiment of the invention in which a monitoring chamber is added.
Fig. 14 shows a schematic representation of the monitoring chamber of Fig. 13. viewobtainedby
10 Fig. 15 shows a tuna contour model used to create a two-dimensional model.
fifteen Fig. 16 shows an unequal distribution of flexion along the spine segments of an example tuna model. Fig. 17 shows three schemes of possible tuna contours generated using the example model, considering three cases of global tuna flexion 9 = 15 °, 9 = 30 ° and 8 = 45 '.
twenty Fig. 18 shows a graphical representation of the five parameters of the example model that define tuna.
25 Fig. 19 shows a real example of the tuna detection process by fitting to an example model.
Detailed description of embodiments of the invention
JO In the embodiments described below, a system is used to carry out the biomass count, the size and number of tunas for a zone (1). For this purpose, measuring means (2) composed of two chambers are used to obtain a sequence of stereoscopic images with two sub-sequences, and an acoustic sensor composed of an echo sounder and a transducer, which
Generate an echogram. For use, the measuring means (2) are installed submerged in a known position and facing upwards, so that ventral measurements are made of the tunas that cross the water zone (1). The echogram is used to determine the number of tunas, while images are used to
5 determine your size and, consequently, your weight.
In Fig. 1 an example of visualization can be observed in which they are shown
simultaneously an image captured by one of the cameras in the sequence of
images and a fragment of the echogram for that moment. It can be seen that the
The contrast between the silhouettes of tunas and the bottom despite the fact that due to turbidity comes a point where they cease to be seen clearly. In this exemplary embodiment both the chambers and the transducer are located at the bottom of a fattening cage, and focused in the vertical direction. In this example, the transducer does not have a curtain beam but is a conical or cardioid beam.
Fig. 2 shows a graphic representation of an example of tuna selection according to the invention. In this example, we start from two stereoscopic frames, one of each sub-sequence of images. From two characteristic points (3) of a tuna in a first frame, in this case the tip of the head and the tip of the tail,
20 epipolar lines (4) are drawn on the second stereoscopic frame of the other sub-sequence of images and images are searched in which their two characteristic points (3) are located on said epipolar lines (4). This creates a correspondence between the tuna of the first frame and the second frame, and is considered to be the same. Since the relative positions of the
25 cameras can triangulate the distance that the tuna is and compensate for the size measurement considering that distance. Likewise, the effect of lens distortion is corrected.
According to an exemplary embodiment of the invention, starting from the
30 frames of the sequence of images are selected objects that correspond to tunas. For this, a blob detection technique known in the field of computer image recognition is used. The detection of the zones that correspond to a candidate object is carried out by means of local contrast criteria, since, although the brightness of the bottom and the fish can vary, the bottom is always clearer than the fish. In Fig. 3 the bottom is seen in the upper left. The lower left corner shows the silhouettes of the candidate objects while in the lower right area you can see the areas that
5 correspond to said candidate objects. Next, a filtering is carried out by geometric criteria of shape and size, discarding those objects that are clearly not tuna, or that cannot be properly segmented. In the upper right part of Fig. 3 you can see silhouettes that correspond to those objects in the image that are considered tunas.
10 Another embodiment uses a tuna model to further refine the selection of zones. Fig. 4 shows a simple example of said model, other examples may include fins or other characteristic elements of the tuna silhouette. Thus, Fig. 5a shows how to try to fit a tuna model in the areas
15 corresponding to segmented candidate objects in the previous case. Once the best position for the model is obtained, the adjustment error is calculated. In the case of the example, the model has a plurality of silhouette points, the error measured is the quadratic sum of the distance between the points of the model and those of the silhouette of the candidate object. If this error is too large, the candidate object is discarded.
20 In Figs. 5a to 5c it can be seen how, despite everything, some silhouettes that really correspond to two small (or distant) tunas have been confused with that of a larger tuna. The improvement of these cases is achieved by also improving the tuna model to be used. In this example, the size is derived from the dimensions of the model for those candidate objects that have been considered as tunas.
25 For the embodiment of the example, the distance calculation strategies based on stereoscopic images detailed above are used and tuna sizes are obtained for those objects of the image that have been successfully processed. The weight of each of these selected tunas is then inferred
30 and the average weight is used to calculate the total biomass, the total value being the result of multiplying the average weight by the number of tunas detected by acoustic techniques described below.
In the form of the example an echogram is used as shown in Fig. 6, which is segmented by image processing techniques based on morphological criteria to obtain segmented regions such as that of Fig. 7. For the sake of clarity no Morphological segmentation techniques are detailed given the great diversity that exists in their technique. It goes without saying that criteria for the shape and determination of zones are used through pre-established models of the echogram traces that are expected to be for tunas. That is, it is based on a model of what is the ultrasound "footprint" of a tuna and similar areas are searched for in the echogram, thus obtaining the number of copies. Since the captured tunas 10 usually move between sizes and with expected echo powers, ranges for these TS and size parameters are chosen and those regions of the ultrasound that fall outside these margins are filtered. On the other hand, the data of water temperature, salinity and pH are incorporated to compensate for the echogram and eliminate the effects of the variations that these conditions may
15 provoke.
In the following embodiments of the invention, the procedure is applied to the counting and determination of size and biomass during tuna transfer operations between cages (6, 7), for example, from capture nets to cages of 20 transport, or from transport cages to fattening farms. For this, a support structure (5) is used that joins both cages (6, 7) and that has windows (8, 9) passing through said cages (6, 7). In this way, when opening the windows, a tuna passage area is enabled between the first window (8) and the second window (9). In the example of Fig. 8, axes representing a first direction (101), which coincides with the direction of passage of the tunas, a second direction (102) horizontal and perpendicular to the first direction (101) have been represented. , as well as the vertical direction (100). Although only shown in this figure, the diagram is equivalent for Figs. 9 to 13. In these examples the measurement interval is the time in which tunas pass from the first cage (6) to the second cage (30). 7). In the examples, cameras with 6mm focal length lenses and a 0.50 aperture curtain transducer in the first direction (101) and 490 aperture in the second direction (102), placed in the vertical direction (100), are used. The passage area has dimensions of 10m in the vertical direction (100) by 10m in the
second direction (102), and the measuring means are located at a depth of about 16m, whereby they are 6m from said passage zone. In this way the 49 ° opening does not cover the entire passage area, which for that distance would require an angle of about 80 °. However, there are two reasons for this selection, apart from 5 the difficulty of finding commercial transducers with such extreme beam widths: the first is that, when talking about beamwidth, it refers to a drop of -3dB, but really still receiving signal outside that width, so in extreme cases, it could still be detected. The second reason is that increasing the depth affects the image component, so that the images
10 would lose sharpness, in particular, for a beam width of 49 °, the distance to the corridor should be about 11m. In addition to the above, the area that is outside the main beam is located at the corners of the lower end, where it is more unlikely that the fish will pass, given their observed behavior.
15 To ensure that, at the speeds at which tunas pass from one cage to another, at least two samples, sometimes called cuts, are obtained from the fish, in the exemplary embodiments, the acoustic sensor emits bursts of pulses of 64 microseconds per burst, repeated at a frequency of 10Hz. Higher frequencies are preferable to be able to make more cuts of the fish, although the cost
20 of the teams may be higher. The maximum for these example configurations is around 46 Hz.
In an exemplary embodiment shown in Fig. 8 a support structure is used
(5) with a rhombus-shaped profile, so that the upper vertex is the point of
25 junction between the two cages (6, 7). The chambers and the transducer are centered in the interior vertex, pointing in a vertical direction (100), so that the water zone (1) corresponds to the tuna passage zone within the structure (5). In Fig. 8 the acoustic beams and the cone of vision of the cameras have been schematically represented. The same has been represented for the rest of the figures in Fig. 9 a
30 12.
In another exemplary embodiment shown in Fig. 9 and 10, a support structure (5) with a pentagonal shaped profile is used, so that the upper vertex is the
5 junction point between the two cages (6, 7). The cameras and the transducer are positioned centered on the inner flat base of the structure (5). In this case the transducer always points in the vertical direction (100), while the cameras point either towards the second window (9) in the case of Fig. 9, or towards the first window (8) in the case of the Fig. 10. The objective of this angulation with respect to the vertical direction (100) is to avoid the shadows of the cage's own floating structures (6, 7).
10 Another embodiment shown in Fig. 11 comprises a support structure (5) with a rectangular profile and in which all measuring means point in a vertical direction (100), centered on the bottom of the structure (5).
15 20 Finally, Fig. 12 shows an alternative embodiment in which the measuring means are located outside the internal passage area of the structure (5) to avoid shadows. In the exemplified case it is located in the second cage (7), in front of the second window (9), but it could also be in the first cage (6) in front of the first window (9). Finally, Fig. 13 shows an embodiment in which a control chamber for visual inspection of tuna transfer operations has been added. This camera is independent of the measuring means and is located pointing to one of the windows (8, 9), preferably to the first window (8). Fig. 14 shows an example of the field of view of the control camera.
2S Description of an embodiment of the tuna model It is an objective of the invention the automatic segmentation of the images obtained by the capture means, as well as the extraction of the size characteristics of the tunas detected therein.
30 For this, a tuna body model is proposed. In particular it is a two-dimensional model of the silhouette from a ventral view, which has the advantage of simpler modeling than in three dimensions, and that reduces the necessary computational cost. The model is also deformable, to adapt to the
forms that tuna can take when swimming and flexing, and
allow precise size values to be obtained even if the silhouette
obtained not from the fish in rectilinear position.
5 Thus, the objective is to obtain a deformable 2D model that takes into account the curvature of the tuna body while swimming. The actual shape of the bluefin tuna body shown in Fig. 15 And some data on the kinematics have been studied experimentally to transfer this knowledge to the model. The modeled form corresponds to a view of the tuna from below, since this perspective of the body of the
10 tuna is quite stable despite its movements and turns when swimming. The developed model takes into account the representative points of the tuna silhouette and its swimming gesture. In Fig. 15 you can see some reference points along the spine, and some extreme points that correspond to the tuna contour. SI represents a unit of measure used as a reference in model a
15 scale factor mode.
In this way, the example model shown in the figures comprises a set K of 39 reference points (also called landmark) for the contour of the tuna body, with 19 reference points to represent each side of the tuna body and 20 other snout point. From the longitudinal axis of the fish that would represent its spine and that goes from the tip of the mouth to the end of the caudal peduncle, sixteen equidistant sections are obtained to define reference positions in the spine and find the possible limits of the body of the Fish and model them. All this is based on the assumption that the body of the tuna under this
25 perspective is practically symmetrical with respect to your spine, when it is not flexed. These vertebrae are labeled Vo and V19 in Fig. 15.
From 20 referenced vertebrae, a set of coefficients that allow us to estimate the ends of the body edge have been obtained experimentally
30 of the tuna on both sides of the spine with respect to these vertebrae. These coefficients relate the position of the vertebrae and the edges of the tuna's body, taking the following values for the case of a non-flexed tuna: 0.7, 1.15, 1.35, 1.55, 1.65, 2.0, 2.5, 2.8, 3.15, 1.7, 1.55, 1.35, 1.1, 0.9, 0.7, 0.55, 0.45, 0.5, 0.75]
On the other hand, during the swimming movement, tunas perform a global flex
e, defined as the angle formed between the first and the last vertebra. It has been
observed that this global flexion e is not distributed evenly throughout the
5 tuna body length. It has been observed that each vertebra contributes differently to this global flexion e. The calculation of the flexion of each vertebra is based on the expressions:
j; 19
"
"d8 1-'e · I! lo¡ ::: 16; 'ti v¡ E v;
L
¡= O ¡; or
10 The values of the coefficients l!, E¡ represent the contribution of the flexion of each vertebra to the global flexion. After a study of the flexion of the individuals, the following values were obtained for the head-to-tail flexion coefficients:
~ 8 = [O, O, 0.64, 0.64, 0.48, 0.48, 0.0, 0.0, 0.0, 0.0, 0.48, 0.64, 0.8, 0.96, 1.12, 1.28, 1.44, 1.92, 2.4, 2.72]
The values of these coefficients show high flexion in the vertebrae close to the tail and very little flexion in the head. In this case the partial flexion of each vertebra would be obtained with the expression:
"
I! J. (J¡: = 16; 'ti v¡ E v; ¡; or
The curvature of the tuna's spine is calculated with the flexion of each vertebra
20 and the reference points for the contour of a detected tuna are obtained. Taking into account the coordinates of each vertebra Vi ::: (xj, yj) you get the
coordinates of the landmarks k¡ ::: (x ~, y ~) using the following expressions:
X¡k: = X (-S, C¡ sin «(J¡); Yt == YI + S¡c¡ cos (8¡); 1 = 0, ..., 19
4 == XI_ 19 + s, C¡_19 sin ((J¡_19); Yt == YI-I <J -SI C¡_19 COS ((J¡_19); i = 20, ..., 39
In Fig. 17 three possible forms of tuna can be observed using the example model for three cases of global flexion 9 = 15 °, 9 = 30 ° and 9 = 45 °.
5 In addition to the flexural capacity by means of parameters 9, the model introduces four parameters to be adaptable to translation and rotation without changing its structural characteristics. Fig. 18 shows these parameters.
10 15 Therefore, the example tuna model, represented by the letter M, is defined by a vector of five parameters M = [sx. sy.l. a, 8] where: sx and Sy are translation parameters that give the location in the muzzle image; l is the length of the spine (l = 16 SI) 'given by the scale factor; a indicates the rigid rotation of the model, defined as the angle of the fish's head in relation to the horizontal axis; and 8 is the angle of global flexion of the spine.
20 25 For this example, in the tuna detection process a phase of adjustment to the model of each of the images of the stereoscopic sequence is performed. The objective of the adaptation process is to obtain the optimum values for the 2D model parameters that best fit a candidate blob to be tuna. It involves adjusting the model to minimize a measure of discrepancy of the model to the image based on the limits of the candidate in the image. This adjustment should minimize the measure of discrepancy between the edges of the model and the edges of the segmented blob or segmented tuna in the image. The model parameters are initialized taking into account blob parameters obtained after segmentation and are adjusted and recalculated until a better fit is no longer achieved.
JO An adjustment error index (FEI) has been defined based on the quadratic distance that occurs between the characteristic points of the model and those of the segmented blob. Figs. 5a to 5c illustrate the adjustment process carried out and Fig. 19 shows a real example. In this case, a tuna model without lateral fins is used, which includes peduncle modeling. In addition the image represents the process for the pair of
stereoscopic images and a tuna without concealments or overlaps with others has been automatically selected to obtain reliable biometric measurements.
The person skilled in the art will understand that from this model others can be derived
5 more detailed, for example, with a greater number of vertebrae, adding a modeling of the peduncle of the fish, or even models with non-constant distribution of the vertebrae. Thus it is possible to obtain greater precision in the calculated biometric measurements that provide the length of the fish taking into account its flexion and the width of the fish.

权利要求:
Claims (36)
[1]
5 1.-Procedure for the determination of tuna biomass in a water zone (1), characterized in that measuring means (2) comprising an acoustic sensor and means are submerged under said water zone (1). for capturing images, said measuring means (2) being in known positions and directed in an upward direction, the process comprising the steps
10 following: during a measurement interval in which some tunas cross said water zone (1):
said acoustic sensor performs acoustic measurements by emitting acoustic signals and receiving echoes, thus generating an echogram; Y
15 said image capture means make image captures, generating an image sequence containing images of said tunas;
determining a value of the number of tunas that have crossed said water zone (1) during said measurement interval using said echogram;
20 determining a tuna weight value using said sequence of images; determining a biomass value of tunas that have crossed said water zone (1) during said measurement interval from said value of the number of tunas and said tuna weight value.
Method according to claim 1, characterized in that determining said tuna weight value comprises: selecting tuna from said sequence of images; determine a size value calculation for each of said selected tunas;
30 determining an individual weight value calculation for each of said tunas selected from said size value; and determining said tuna weight value as the average value of said individual weight value for each of said selected tunas;
wherein said biomass value is determined as the multiplication of said tuna number value by said tuna weight value.
[3]
3. Method according to claim 2, characterized in that said means of
The image capture comprises two optical sensors located at a relative position between them, so that said image sequence comprises a stereoscopic sequence comprising a first sub-sequence of images and a second sub-sequence of images, both sub-sequences being synchronized; wherein said procedure comprises the additional steps of:
10 selecting tuna present simultaneously in said first sub-sequence of images and in said second sub-sequence of images; determining a distance value calculation for said selected tunas, using the values of said relative position between said first optical sensor and said second optical sensor; Y
15 modify said size value for each of the selected tunas using said distance value;
[4]
4. Method according to claim 3, characterized in that selecting tuna present simultaneously in said first sub-sequence of images and in
Said second sub-sequence of images comprises selecting two characteristic points (3) of each tuna selected in one of said first sub-sequence and said second sub-sequence, said two characteristic points being
(3) preferably the tip of the head and the tip of the tail, and projecting said two points on the other between said first sub-sequence and said second sub
25 sequence, through epipolar lines (4); so that those tunas with said two characteristic points (3) are selected in one of said first sub-sequence and said second sub-sequence that correspond to another two characteristic points (3) located in said epipolar lines (4) in the other between said first sub-sequence and said second sub-sequence.
Method according to claim 4, characterized in that determining a distance value calculation and modifying said size value for said selected tunas comprises the steps of:
correct distortion effects of the lenses in said sequence of images; obtaining the distance value by triangulation of each pair of points of said sequence of images, wherein said pair of points comprises a point in one of said first sub-sequence and said second sub-sequence, and a corresponding point in the another between said first sub-sequence and said second sub-sequence; and modifying said size value for each selected tuna based on said distance value of said pairs of points corresponding to said selected tuna.
[6]
6. Method according to any of claims 2 to 5, characterized in that said sequence of images comprises a plurality of frames and in which selecting tunas from said sequence of images comprises the following steps:
segment said frames to obtain first candidate objects; filter said first candidate objects by their geometric characteristics, obtaining a few second objects that are a subset of said first objects;
[7]
7. Method according to claim 6, characterized in that selecting
tuna of said sequence of images comprises the additional step of: adjusting a tuna image model to said second objects, so that if for an object the adjustment error exceeds an error threshold value, said object is discarded, while in otherwise, it is considered that said object is a selected tuna and an adjusted model is obtained;
and wherein determining a size value calculation for each of said selected tunas comprises obtaining said size value from said adjusted model of said selected tuna.
[8]
8. Method according to claim 7, characterized in that said tuna image model comprises a segment model in which points, called vertebrae, Vi, are used, located on the tuna axis, preferably 20 points, numbered from the tuna. vertebra O (vo) corresponding to the tip of the head, to vertebra 19 (V19) corresponding to the end of the tail.
[9]
9. Method according to claim 8, characterized in that a
5 estimation of the tuna contour relating each of these vertebrae, Vi, with aestimated point of the contour, k¡, preferably by means of coefficients thatrelate the position of each of said vertebrae Vi with the distance to the vertebraV¡ of each of said estimated points of the contour k¡, preferablyfollowing coefficients:
CU = [O, 0.7, 1.15, 1.35, 1.55, 1.65, 2.0, 2.5, 2.8, 3.15, 1.7, 1.55, 1.35, 1.1, 0.9, 0.7, 0.55, 0.45, 0.5, 0.75]
10 where the values are sequentially ordered from Vo to V19.
[10]
10. Method according to any of claims 8 to 9, characterized by
that said model assigns a different contribution value of each vertebra given a
global flexion, e, preferably following the following formulation:
i = 19
"
IdO, = O; L! J.fJi = 16; 'r / Vi E V;
i = O; = 0
where you represent the global flexion and the values of the coefficients!
contribution of the flexion of each vertebra V¡ to the global flexion e, preferably taking the following values: "8 = [O, O, 0.64, 0.64, 0.48, 0.48, 0.0, 0.0, 0.0, 0.0, 0.48, 0.64, 0.8 , 0.96, 1.12, 1.28, 1.44, 1.92, 2.4, 2.72]
20 where the values are sequentially ordered from Vo to V19.
[11 ]
11. Method according to any of claims 8 to 10, characterized in that said model also comprises translation and rotation parameters.
12. Method according to any of claims 1 to 11, characterized in that determining said value of the number of tunas that have crossed said area of water (1) during said measurement interval using said echogram comprises the
stages of: segmenting regions of said echogram by applying mathematical morphology criteria, considering that said echogram is an image;
5 obtain segmented regions; and determine said value of the number of tunas as the value of the number of said segmented regions.
[13]
13. Method according to claim 12, characterized in that said criteria
10 of mathematical morphology comprise ultrasound trace models for tunas, so that said regions are segmented by morphological similarity with said trace models.
[14]
14. Method according to any of claims 12 or 13, characterized in that control parameters are chosen that comprise one or more of the list
consisting of:target strenght margins, TS; Ytuna size margins;
where by segmenting regions of said echogram, those regions corresponding to values outside said control parameters are discarded.
[15]
15. Method according to any of claims 12 to 14, characterized in that it comprises the additional step of processing said echogram to compensate for an effect on said echogram caused by any of the parameters included in
25 the list consisting of: water temperature; water salinity; water pH;
16. Method according to any of claims 1 to 15, characterized in that a support structure (5) is provided that joins a first cage (6) with a second cage (7), wherein said support structure (5) presents a first window
(8) passing that communicates with said first cage (6) and a second window (9) of
step that communicates with said second cage (7), defining a passage zone for tunas between said first window (8) and said second window (9), a direction of passage that goes horizontally from said first cage (6) to said second cage (7), thus defining a first direction (101) that coincides with said step direction 5, and a second direction (102) that is horizontal and perpendicular to said first direction (101); wherein said structure has a first profile looking in said first direction (101), and a second profile looking in said second direction (102); and wherein said measuring range comprises a time in which a plurality of tunas moves from said first cage (6) to said second cage
10 (7) in said direction of passage.
[17]
17. Method according to claim 16, characterized in that said sensor
acoustic features a detection beam in the form of a curtain with a wide side and a
narrow side, said sensor arranged so that said wide side extends in
Said second direction (102) having a second beam width such that it covers all said water zone (1), and said narrow side extends in said first direction (101) presenting a first beam width less than 5 °, preferably less than 2 °, more preferably between 1 ° and 0.5 °.
18. Method according to claim 17, characterized in that said acoustic sensor emits bursts of pulses with a burst duration of less than 100 microseconds, preferably less than 50 microseconds.
[19]
19. Method according to any of claims 17 or 18, characterized by
25 that said acoustic sensor emits pulses that are repeated with a frequency greater than or equal to 10Hz, preferably between 30 and 40 Hz.
[20]
20. Method according to any of claims 17 to 19, characterized in that said acoustic sensor comprises an echo sounder and a wide beam transducer,
30 wherein said second beam width is between 40 ° and 80 °, preferably between 45 ° and 55 °, more preferably 49 °.
[21]
21. Method according to any of claims 17 to 20, characterized in that said measuring means (2) are positioned on the inner bottom of said support structure (5), centered with respect to said second direction (102), and in the that said acoustic sensor is positioned directed in the vertical direction (100), so that said
5 water zone (1) is included in said passage zone.
[22]
22. Method according to claim 21, characterized in that said measuring means (2) are positioned centered with respect to said first direction (101).
23. Method according to any of claims 21 or 22, characterized in that said second profile of said support structure (5) has a rhombus shape, with an upper vertex and a lower vertex, and wherein said means of measure (2) are positioned in said lower vertex and directed in the vertical direction (100).
24. Method according to any of claims 21 or 22, characterized in that said support structure (5) has a second pentagon-shaped profile with an upper vertex and a lower flat base, wherein said measuring means
(2) are positioned on said flat base and in which said image capture means
they are positioned directed towards said first window (8) or said second window (9).
[25]
25. Method according to any of claims 21 or 22, characterized in that said second profile of said support structure (5) has a rectangular shape, with a flat bottom and top part, and in which all
Said measuring means (2) are positioned in said lower part and directed in a vertical direction (100).
[26]
26. Method according to any of claims 16 to 20, characterized in that said measuring means (2) are arranged outside said support structure (5) and directed in a vertical direction (100), positioned in said first cage
(6) so that said water zone (1) is located in front of said first window (8), or in said second cage (7) so that said water zone (1) is located in front of said second window (9).
[27]
27.-System for the determination of tuna biomass in a water zone (1), characterized in that it comprises measuring means (2) comprising an acoustic sensor and image capture means, configured to be
5 arranged submerged under said water zone (1), in known positions and directed in an upward direction, during which during a measurement interval in which tunas cross said water zone (1):
said acoustic sensor is configured to perform acoustic measurements by emitting acoustic signals and receiving echoes, thus generating a
10 echogram; and said image capture means are configured to perform image captures, generating a sequence of images containing images of said tunas;
said system also comprising analysis means configured to:
15 determining a value of the number of tunas that have crossed said water zone (1) during said measurement interval using said echogram; determining a tuna weight value using said sequence of images; determining a biomass value of tunas that have crossed said water zone (1) during said measurement interval from said value of the number of
20 tunas and of said tuna weight value.
[28]
28. System according to claim 27, characterized in that determining said tuna weight value comprises: selecting tuna from said sequence of images;
25 determine a size value calculation for each of said selected tunas; determine an individual weight value calculation for each of said tunas selected from said size value; and determine said tuna weight value as the average value of said value of
30 individual weight for each of these selected tunas; wherein said biomass value is determined as the multiplication of said tuna number value by said tuna weight value.
[29]
29. System according to claim 28, characterized in that said image capture means comprise two optical sensors located at a relative position between them, such that said image sequence comprises a stereoscopic sequence comprising a first sub-sequence of images Y
5 a second sub-sequence of images, both sub-sequences being synchronized; wherein said analysis means are configured to perform the additional steps of:
selecting tuna present simultaneously in said first sub-sequence of images and in said second sub-sequence of images;
10 determining a distance value calculation for said selected tunas, using the values of said relative position between said first optical sensor and said second optical sensor; and modify said size value for each of the selected tunas using said distance value;
[30]
30. System according to claim 29, characterized in that selecting tuna present simultaneously in said first sub-sequence of images and in said second sub-sequence of images comprises selecting two characteristic points (3) of each tuna selected in one of said first sub-sequence and said second sub-sequence, said two characteristic points being
(3) preferably the tip of the head and the tip of the tail, and projecting said two points on the other between said first sub-sequence and said second sub-sequence, through epipolar lines (4); so that those tunas with said two characteristic points (3) are selected in one of said
25 first sub-sequence and said second sub-sequence corresponding to two other characteristic points (3) located in said epipolar lines (4) in the other between said first sub-sequence and said second sub-sequence.
[31 ]
31. System according to claim 30, characterized in that determining a distance value calculation 30 and modifying said size value for said tunas
selected includes the steps of: correcting distortion effects of the lenses in said sequence of images; obtaining the distance value by triangulation of each pair of points of said sequence of images, wherein said pair of points comprises a point in one of said first sub-sequence and said second sub-sequence, and a corresponding point in the another one between said first sub-sequence and said
5 second sub-sequence; and modifying said size value for each selected tuna based on said distance value of said pairs of points corresponding to said selected tuna.
32. System according to any one of claims 28 to 31, characterized in that said sequence of images comprises a plurality of frames and in which selecting tunas from said sequence of images comprises the following steps:
segment said frames to obtain first candidate objects;
15 filter said first candidate objects by their geometric characteristics, obtaining second objects that are a subset of said first objects;
[33]
33.-System according to claim 32, characterized in that selecting
20 tunas of said sequence of images comprise the additional step of: adjusting a tuna image model to said second objects, so that if for an object the adjustment error exceeds an error threshold value, said object is discarded, while otherwise, it is considered that said object is a selected tuna and an adjusted model is obtained;
25 and wherein determining a size value calculation for each of said selected tunas comprises obtaining said size value from said adjusted model of said selected tuna.
[34]
34.-System according to any of claims 27 to 33, characterized in that said value of the number of tunas that have crossed said water zone is determined
(1) during said measurement interval using said echogram it comprises the steps of: segmenting regions of said echogram by applying mathematical morphology criteria, considering that said echogram is an image; obtain segmented regions; and determine said value of the number of tunas as the value of the number of said
5 segmented regions.
[35]
35. System according to claim 34, characterized in that said criteria of
Mathematical morphology includes some ultrasound trace models for
tunas, so that these regions are segmented by morphological similarity with
10 said trace models.
[36]
36. System according to any of claims 34 or 35, characterized in that said analysis means are configured to allow the choice of control parameters comprising in one or more of the list consisting of:
15 margins of target strenght, T5; Y
tuna size margins; where said analysis means are configured so that when segmenting regions of said echogram, discard those regions that correspond to values outside said control parameters.
37. System according to any of claims 34 to 36, characterized in that said analysis means are additionally configured to process said echogram to compensate for an effect on said echogram caused by any of the parameters included in the list consisting of:
25 water temperature; water salinity; water pH;
[38]
38. System according to any of claims 27 to 37, characterized in that
30 further comprises a support structure (5) configured to be arranged in a use position in which it joins a first cage (6) with a second cage (7), where said support structure (5) has a first window ( 8) of passage that communicates with said first cage (6) and a second window (9) of passage that
communicates with said second cage (7), defining a passage area for tunas between said first window (8) and said second window (9), a passage direction that goes horizontally from said first cage (6) to said second cage ( 7), thus defining a first direction (101) that coincides with said direction of passage, and a second direction (102) that is horizontal and perpendicular to said first direction (101); wherein said structure has a first profile looking in said first direction (101), and a second profile looking in said second direction (102); and wherein said system is configured so that the measurement range comprises a time in which a plurality of tunas travels from said first cage (6) to
10 said second cage (7) in said direction of passage.
[39]
39. System according to claim 38, characterized in that said acoustic sensor has a detection beam in the form of a curtain with a wide side and a narrow side, so that in said position of use said sensor is arranged so
15 that said wide side extends in said second direction (102) having a second beam width such that it covers all said water zone (1), and said narrow side extends in said first direction (101) presenting a first width of beam less than 5 °, preferably less than 2 °, more preferably between 1 ° and 0.5 °.
40. System according to claim 39, characterized in that said acoustic sensor is configured to emit bursts of pulses having a burst duration of less than 100 microseconds, preferably less than 50 microseconds.
[41]
41.-System according to any of claims 39 or 40, characterized in that
25 said acoustic sensor is configured to emit pulses having a repetition frequency greater than or equal to 10 Hz, preferably between 30 and 40 Hz.
[42]
42. System according to any of claims 39 to 41, characterized in that said acoustic sensor comprises an echo sounder and a wide beam transducer, in the
30 that said second beam width is between 40 ° and 80 °, preferably between 45 ° and 55 °, more preferably 49 °.
[43]
43.-System according to any of claims 39 to 42, characterized in that in said position of use said measuring means (2) are positioned in the inner bottom of said support structure (5), centered with respect to said second direction (102), and wherein said acoustic sensor is positioned directed at
5 vertical direction (100), so that said water zone (1) is comprised in said passage zone.
[44]
44. System according to claim 43, characterized in that in said position of use said measuring means (2) are positioned centered with respect to said first direction (101).
[45]
45. System according to any of claims 43 or 44, characterized in that
said second profile of said support structure (5) has a diamond shape,
with an upper vertex and a lower vertex, and in which in said position of use
15 said measuring means (2) are positioned in said lower vertex, and directed in the vertical direction (100).
[46]
46. System according to any of claims 43 or 44, characterized in that said support structure (5) has a second pentagon-shaped profile with
20 an upper vertex and a lower flat base, in which in said position of use said measuring means (2) are positioned in said flat base and in which said image capture means are positioned directed in the direction of said first window (8) or said second window (9).
47. System according to any of claims 43 or 44, characterized in that said second profile of said support structure (5) has a rectangular shape, with a flat bottom and a top part, and in which in said position of use all said measuring means (2) are positioned in said lower part and directed in a vertical direction (100).
30. System according to any of claims 38 to 42, characterized in that in said position of use said measuring means (2) are positioned outside said support structure (5) and directed in a vertical direction (100 ),
positioned in said first cage (6) so that said water zone (1) is located in front of said first window (8), or in said second cage (7) so that said water zone (1) is located in front of said second window (9).

类似技术:

---

公开号 | 公开日 | 专利标题

---

ES2665583B1|2019-03-04|PROCEDURE FOR THE DETERMINATION OF TUNA BIOMASS IN A WATER AREA AND A CORRESPONDING SYSTEM

---

CN105654549B|2018-05-15|Underwater 3 D reconstructing device and method based on structured light technique and photometric stereo

---

CN103971406B|2017-12-08|Submarine target three-dimensional rebuilding method based on line-structured light

---

Pitcher1973|The three-dimensional structure of schools in the minnow, Phoxinus phoxinus |

---

Gershikov et al.2013|Horizon line detection in marine images: which method to choose?

---

CN107219529B|2019-08-09|A kind of acquisition methods and system of high-precision seafloor topography figure

---

Bagnitsky et al.2011|Side scan sonar using for underwater cables & pipelines tracking by means of AUV

---

GB2539495A|2016-12-21|Improvements relating to time-of-flight cameras

---

Doi2013|Further development of whale sighting theory

---

de Jesus et al.2015|Reconstruction accuracy assessment of surface and underwater 3D motion analysis: a new approach

---

Hurtós et al.2010|Calibration of optical camera coupled to acoustic multibeam for underwater 3D scene reconstruction

---

WO2020046524A1|2020-03-05|Automatic feed pellet monitoring based on camera footage in an aquaculture environment

---

CN105741284B|2018-10-26|A kind of multi-beam Forward-looking Sonar object detection method

---

CN103226828A|2013-07-31|Image registration method of acoustic and visual three-dimensional imaging with underwater vehicle

---

Ben-Simon et al.2009|Measuring and tracking eye movements of a behaving archer fish by real-time stereo vision

---

Westman et al.2019|Wide aperture imaging sonar reconstruction using generative models

---

CN110231021A|2019-09-13|Ripple sensor, ripple method for reconstructing and its application

---

Pérez et al.2018|Automatic measurement of fish size using stereo vision

---

Burguera et al.2014|Intensity correction of side-scan sonar images

---

KR101566337B1|2015-11-06|Artificial landmark for facilitating identification by underwater imaging sonar and image processing

---

CN106096617A|2016-11-09|A kind of based on the panorama sea horizon extracting method improving active contour model

---

Spears et al.2014|Determining underwater vehicle movement from sonar data in relatively featureless seafloor tracking missions

---

Casagrande2013|Real-time featureless visual odometry for sea floor imaging with a Lagrangian float

---

CN104536005B|2017-04-12|Multi-beam side-scan sonar slant-distance distortion elimination method based on blind area correction

---

Burguera et al.2015|Building High Resolution Maps of Large Subsea Areas using Side-Scan Sonar

同族专利:

---

公开号 | 公开日

---

MA51648A|2018-05-02|

---

ES2665583B1|2019-03-04|

---

EP3316220A1|2018-05-02|

---

EP3316220B1|2022-01-05|

引用文献:

---

公开号 | 申请日 | 公开日 | 申请人 | 专利标题

---

JPH02113374A|1988-10-21|1990-04-25|Joho Seigyo Syst Kk|Detecting device for underwater floating body|

---

US5377163A|1993-11-01|1994-12-27|Simpson; Patrick K.|Active broadband acoustic method and apparatus for identifying aquatic life|

---

US5692064A|1993-11-01|1997-11-25|Hitachi, Ltd.|Method and apparatus for counting underwater objects using an ultrasonic wave|

---

GB2522302A|2013-10-31|2015-07-22|Furuno Electric Co|Size-and-type determining device, underwater detecting apparatus and method of determining size and type|CN112384768A|2018-05-04|2021-02-19|艾克斯波特西溶液公司|Scale for determining the weight of a biological body|US20110292201A1|2010-05-27|2011-12-01|Waylon Westphal|Underwater scouting camera|DE102018215051B3|2018-09-05|2019-10-10|Atlas Maridan Aps|Fish observation device|

---

DE102018217164B4|2018-10-08|2022-01-13|GEOMAR Helmholtz Centre for Ocean Research Kiel|Method and system for data analysis|

---

EP3754374A1|2019-06-17|2020-12-23|Furuno Electric Co., Ltd.|Fish weight measuring apparatus|

---

WO2021059143A2|2019-09-27|2021-04-01|Agam Aquaculture Ltd.|Fish management system and method|

---

US20220000079A1|2020-07-06|2022-01-06|Ecto, Inc.|Acoustics augmentation for monocular depth estimation|

法律状态:
2019-03-04| FG2A| Definitive protection|Ref document number: 2665583 Country of ref document: ES Kind code of ref document: B1 Effective date: 20190304 |

优先权:

---

申请号 | 申请日 | 专利标题

---

ES201631379A|ES2665583B1|2016-10-26|2016-10-26|PROCEDURE FOR THE DETERMINATION OF TUNA BIOMASS IN A WATER AREA AND A CORRESPONDING SYSTEM|ES201631379A| ES2665583B1|2016-10-26|2016-10-26|PROCEDURE FOR THE DETERMINATION OF TUNA BIOMASS IN A WATER AREA AND A CORRESPONDING SYSTEM|

---

EP17197860.4A| EP3316220B1|2016-10-26|2017-10-23|Method for determining tuna biomass in a water zone and corresponding system|

---

MA051648A| MA51648A|2016-10-26|2017-10-23|METHOD FOR DETERMINING TUNA BIOMASS IN A WATER ZONE AND CORRESPONDING SYSTEM|