• 专利附录
  • 专利说明
  • 权利要求
  • 类似技术
  • 同族专利
  • 引用文献
  • 法律状态
  • 优先权
    • 专利摘要:
    A gestural recognition arrangement (1) comprises two spaced-apart units (3, 4) for detecting two-dimensional thermal radiation-based image information, respectively, means (5) for merging the image information acquired by the two thermal imaging units (3, 4) in order to obtain three-dimensional image information, and means (6) for evaluating a plurality of three-dimensional image information determined temporally successively for detecting a gesture.
    公开号:AT15034U1
    申请号:TGM172/2014U
    申请日:2014-04-23
    公开日:2016-11-15
    发明作者:
    申请人:Zumtobel Lighting Gmbh；Zeidler Gerd；
    IPC主号:
    专利说明:

    description
    ARRANGEMENT AND METHOD FOR DETECTING GESTURES
    The present invention relates to an arrangement with the aid of gestures can be detected to generate, for example, control information for driving a lamp or a similar device. Furthermore, the present invention relates to a corresponding method for recognizing gestures.
    The setting of devices of the lighting technology can be done in different ways. Known are, for example, mounted on a wall of a room operating devices, which generate control information by a corresponding touch or manual operation of the individual controls. A classic control, for example, a so-called dimmer, which can be pressed on the one hand to turn the light on and off. In addition, for example, it can be rotated or shifted to vary the brightness. More modern input devices are based on the use of so-called touchscreens, which generate control information by touching corresponding surface areas and, moreover, also offer the possibility of displaying information on the screen itself with regard to the operating state of the luminaire or, in general, the lighting. Communication with the devices to be controlled can take place both via cable and wirelessly.
    Another way to control devices is the so-called gesture control, which has become increasingly important in recent years. This is a non-contact method in which movements of a person, who performs these in particular with their hands, are recognized and assigned to corresponding control commands. Such contactless control of devices is particularly advantageous if an immediate touch of controls is not desired or not possible, which may be the case, for example, for reasons of hygiene. In addition, of course, very different movements or gestures can be performed, so that the control options are significantly expanded. A prerequisite for this, however, is that the gestures of the person are recognized error-free and reliable.
    The recently developed technologies are becoming more mature. Especially in the field of game consoles gesture control concepts are already widely used, so that this technology prevails more and more. Accordingly, it would also be interesting for lighting or lighting control, although in this context there still exist some problems which have not yet been completely solved. For a mass-suitable operation, for example in the lighting control, the corresponding sensor system for detecting the gestures should be kept as simple and cost-effective as possible in order to reduce the corresponding expense. However, this usually affects the recognition performance of the system. Often, therefore, the detection of unique and simple gestures such as so-called wiping movements can be realized by means of simpler systems.
    By contrast, more complex gestures, such as rotational movements, which could be used, for example, to adjust the brightness, are either not recognized or at most with a significant delay. As a result, however, the potential applications of such gesture control systems in lighting technology are significantly reduced.
    Various sensor systems for gesture control are known from the prior art, which have the disadvantages described above. A known system is, for example, the so-called HALIOS system or the system GestIC, which each have a relatively simple structure, but then have the above problems, ie, that only a few gestures can be distinguished from each other and recognizing more complex gestures if necessary with a time delay. More sophisticated sensor systems are also known, which then allow a much better gesture recognition. These could be, for example, the so-called Kinect camera, which is used in the field of game consoles, or systems based on so-called time-of-flight cameras. As already mentioned, however, these systems are so elaborately designed that the associated costs exclude their use in lighting control.
    The present invention is therefore based on the object to provide a novel solution for detecting gestures available, which on the one hand is simple and therefore inexpensive to implement, on the other hand, the requirements in particular with regard to the control of lights or lighting systems.
    The object is achieved by an arrangement for detecting gestures with the features of claim 1 and by a method for detecting gestures according to claim 7. Advantageous developments of the invention are the subject of the dependent claims.
    Basically, the concept according to the invention is based on the idea of detecting thermal images for the detection of gestures, combining them into three-dimensional information in the context of a so-called stereo reconstruction, and recognizing gestures by evaluating successively acquired three-dimensional image information.
    According to the invention therefore an arrangement for detecting gestures is proposed, which has two spaced-apart units for detecting each of a two-dimensional, based on thermal radiation image information. Furthermore, means are provided for merging the image information acquired by the two thermal image acquisition units in order to obtain three-dimensional image information, and means for evaluating a plurality of three-dimensional image information determined temporally successively to detect a gesture.
    Also, a method for recognizing gestures is proposed which comprises the following steps: detecting two-dimensional thermal radiation-based image information from at least two different positions, merging the two-dimensional image information into three-dimensional image information [0014] · Evaluating a plurality of three-dimensional image information determined temporally successively to detect a gesture.
    By combining the principle known per se to record by means of stereo images three-dimensional information, with the evaluation of thermal radiation or infrared radiation is further made possible, despite relatively simple structure very efficient, especially very reliable and fast gestures. The restriction to thermal radiation makes it possible to actually only consider information that has to be used for the recognition of gestures. Thus, by restricting to image information that is within a certain temperature range, a human hand can be easily separated from the rest of the image background. This so-called segmentation of the thermal image is very easy to implement, which ultimately means that the amount of data to be evaluated can be kept low, which in turn has a positive effect on the delay until the recognition of the gesture.
    In addition, due to the restriction to thermal image information, the arrangement can be realized by relatively inexpensive units. For example, the thermal imaging units can be infrared cameras, but so-called thermopile sensors are preferably used. Such thermopile array sensors are very cost effective compared to more complex cameras. Although they have a relatively low resolution of, for example, 8x8 pixels, it has been found that even such a low resolution is sufficient to be able to reliably recognize even more complex gestures. The small number of pixels to be used for the evaluation, of course, again has a positive effect on the speed of processing the data.
    As already mentioned, a segmentation of the thermal image is preferably carried out to optimize the data processing. That is, only those image information are taken into account, which are within a predetermined temperature range, this temperature range corresponds to the usual temperatures of a human hand.
    Finally, therefore, an arrangement is provided with the solution according to the invention, which can be realized compact and inexpensive, but at the same time allows the recognition of gestures within the shortest possible time and with high precision. This also complex control commands can be generated in a comfortable manner, with the help of e.g. an efficient lighting control can be realized. However, the use of the arrangement according to the invention is not limited to control of lights but can be used in a variety of controllable devices used, in particular also air conditioning, ventilation and / or heating devices, blinds, garage doors, home entertainment equipment or the like would.
    The invention will be explained in more detail with reference to the accompanying drawings. 1 shows schematically the structure of an arrangement according to the invention for recognizing gestures and the principle of stereo recording; FIG. 2 schematically shows the combining of the image information acquired by the two thermal imaging units to obtain three-dimensional image information; Figure 3 shows an example of detecting a human hand by means of a on the
    Use of thermopile sensor based arrangement; Figure 4 shows the possibility of using the inventive arrangement to determine the distance of a detected object; and FIG. 5 shows an example for detecting a rotational gesture.
    The schematic structure of an arrangement according to the invention will first be explained with reference to FIG. The procedure for recognizing the gestures will then be explained in detail with reference to the further figures.
    As already mentioned, the principle of gesture recognition according to the present invention is based on the use of a so-called stereo camera, which captures and evaluates thermal image information according to the invention. The illustrated in Figure 1, generally provided with the reference numeral 1 arrangement according to the invention for recognizing gestures therefore initially has a stereo camera 2, which includes two spaced apart thermal imaging units 3 and 4. These two detection units 3 and 4 are arranged in a known, defined distance from each other and each have independent means for image acquisition. As explained in more detail below, each may be separate infrared cameras, but so-called thermopile sensors are preferably used. The image information captured by the two thermal imaging units 3 and 4 initially represent two-dimensional information. These are then combined by an evaluation electronics 5 of the stereo camera 2 to obtain three-dimensional image information, the corresponding principle being shown schematically in FIG.
    FIG. 2 shows the image captured by one of the two image acquisition units 3, 4, respectively, of the two objects shown in FIG. 1, that is to say of the cube and the ball arranged obliquely behind it. The left thermal imaging unit 3 creates from its point of view the image indicated in Figure 2 with I, in which both objects appear to be arranged almost in succession. The right thermal imaging unit 4, however, captures the image II, in which a spatial separation of the two objects from each other is clearly visible. Both images are now merged and combined by the transmitter 5 together, so that ultimately an image information can be obtained, as indicated in Figure 2 with III. In particular, this image information contains a depth information, which is represented in FIG. 2 by the fact that objects lying closer to darker and more distant objects are displayed brighter. By bringing together the originally two-dimensional image information can thus be determined that the distance between the camera 2 to the cube is less than to the underlying ball.
    The described 3D reconstruction of the stereo recordings recorded by the detection units 3 and 4, which is shown schematically in FIG. 2, makes use of the circumstance that the images of the respective units 3, 4 show different viewing angles. Due to the so-called disparity, that is to say the different distance of the objects in the images from the edge of the image and each other, the depth image shown in FIG. 2 can then be generated with the aid of known algorithms, which contains the desired three-dimensional information. Such algorithms for combining stereo recording to produce a depth image are known in the art.
    The captured in this way by the stereo camera 2 images with depth information are then forwarded to a further evaluation unit 6. Several consecutive shots taken in this way can then be evaluated and used to detect gestures. Of course, the units for combining the images for obtaining the depth information as well as for evaluating consecutively produced images for recognizing gestures can hereby be combined. After a gesture has been detected, a control command is then generated based on it and transmitted in the embodiment of Figure 1 to a light or generally to the device to be controlled. The transmission of this command can be done both via a corresponding cable or a control line and wirelessly.
    Before being explained in more detail with reference to the following figures, in which way the three-dimensional image information is then converted into the recognition of gestures, will first be discussed in more detail on the used thermal imaging units 3, 4. As already mentioned, these are units that detect the so-called heat radiation, ie the far-infrared radiation of objects. It may be in this case, therefore, to classic infrared cameras or so-called thermographic cameras, which have a resolution of about 0.1 to 1 megapixel, according to the current state, but are therefore relatively expensive. In addition, such cameras require active cooling, which further increases the cost and associated costs.
    However, so-called thermopile array sensors are preferably used for the two detection units 3, 4. These are units which react exclusively to heat radiation, with a corresponding sensor array being arranged behind an infrared lens. Such thermopile array sensors typically have a very low resolution compared to other image sensors, with commercially available units with a resolution of 4x4 or 8x8 pixels, but in the future also sensors with a higher resolution will be available. Of course, as a result, the resolution of the captured image information is significantly lower than in the aforementioned thermographic camera, but also allows a resolution of at least 8x8 pixels still a sufficiently reliable detection of gestures, as will be shown below.
    In addition, thermopile sensors have serious advantages, since they are at least partially already available as complete units including data interface and a data evaluation can be done with a simpler microcontroller due to the small number of pixels. Due to a different measuring principle and the ability to use in the manufacture of micromechanical systems (MEMS), the manufacturing cost of such a sensor are significantly lower than the cost of the
    Image sensor of a thermography camera. Finally, with a corresponding thermopile sensor no active cooling is required and for reasons of data protection technology, the associated problems are lower, since due to the low image resolution people can not be identified.
    That despite a low resolution of e.g. 8x8 pixels are enough information for recognizing gestures available, can be found in the below discussed figures 3 to 5.
    Figure 3 shows here first schematically determining a depth image of the hand of a user who wants to generate control commands using the inventive arrangement. It should be noted that in this case, only the hand is shown and was dispensed with the representation of other body parts of the person. With regard to the image information determined by the Thermopi-le sensors, however, it should be noted that they can automatically hide background information and the like, since only thermal image information which is usually present in the temperature range of a human hand is recorded anyway within the scope of an image segmentation. This, too, is an advantage of gesture recognition based on thermal image information, since the hiding of disturbing background information can be much easier and more efficient compared to purely optical images.
    Thus, while FIG. 3a shows the human hand in reality-already in the raster of the 8x8 pixel array-the thermal images produced by the two thermal imaging units are shown in FIGS. It can be seen that due to the low sensor resolution of course some details can not be displayed. For example, the spaces between the fingers have more or less disappeared. Nevertheless, the resulting image contains enough information to realize an object tracking, ie a so-called tracking. Here, the fact of the already mentioned segmentation is exploited, namely that the object can be easily separated from the background and the remaining information provide a characteristic form available, which can be used for object recognition. The two thermal images are then combined into the depth representation according to FIG. 3d.
    This form of image analysis allows, in particular, a distance measurement, as FIG. 4 shows. Shown is the depth image of a human hand, which is ultimately captured by the thermal imager, with the hand being closer to the camera in the left-hand area, whereas it is farther away in the right-hand area of FIG. As before, areas closer to the representation are shown darker than more distant areas, whereby it can be seen that the hand in the left area not only appears larger in the finally created 3D image but is actually recognized as being closer to the camera. By temporally successive acquisition of the image information, as shown in Figure 4, for example, a simple so-called tap gesture, that is, a short virtual tap a sensor could be clearly detected. For this, only the corresponding depth information of the image needs to be evaluated over time.
    Finally, FIG. 5 shows that, in spite of the relatively simple construction, even more complex gestures can be recognized. Illustrated in FIG. 5 in the upper region, for example, is a so-called rotational movement performed with one hand, which could be used, for example, for dimming a luminaire or for changing the setting angle of a venetian blind. The hand is moved in accordance with the illustration in the upper area of Figure 5 in a small circle following the circle line, in which case in successive sequences sequential depth images shown in the lower area are detected. Due to the fact that with the help of the segmentation described above, the hand can be recognized in principle without further notice, this rotational movement can be detected after a short time. In comparison to known, previously used units for detecting gestures, a complete execution of the circular movement is not required here.
    It can thus be seen that the creation of several three-dimensional images in succession and the evaluation of them one after the other can unambiguously and efficiently identify motion vectors with the aid of which the type, orientation and movement of a hand can be determined. This allows a correct and in particular timely recognition of gestures.
    Finally, therefore, with the solution according to the invention, a very simply designed arrangement is provided, which despite everything very quickly and efficiently allows the detection of movements of a human hand. As a result, the foundations are created to be able to rely on the principle of gesture control also for driving simple devices such as lights or lighting systems.
    权利要求:
    Claims (11)
    [1]
    claims
    An arrangement (1) for recognizing gestures, comprising: two spaced-apart units (3, 4) for respectively detecting a two-dimensional heat radiation-based image information, and means for combining the two thermal imaging units (3 , 4) acquired image information in order to obtain a three-dimensional image information, • means (6) for evaluating a plurality of three-dimensional image information determined temporally successively for detecting a gesture.
    [2]
    2. Arrangement according to claim 1, characterized in that it is in the thermal imaging units (3, 4) are infrared cameras.
    [3]
    3. Arrangement according to claim 1, characterized in that it is in the thermal image detection units (3, 4) each a thermopile sensor.
    [4]
    4. Arrangement according to claim 3, characterized in that each thermopile sensor has a resolution of 8x8 pixels.
    [5]
    5. Arrangement according to one of the preceding claims, characterized in that the thermal imaging units (3, 4) and / or the means (5) for merging the image information captured by the two thermal imaging units (3, 4) are designed to only to consider information within a given temperature range.
    [6]
    6. Control device for controlling lights, blinds and / or air conditioners, comprising an arrangement (1) for detecting gestures according to one of the preceding claims and means for creating and transmitting control commands based on the recognized gestures.
    [7]
    7. A method for recognizing gestures, comprising the following steps: a) detecting two-dimensional thermal radiation-based image information from at least two different positions, b) merging the two-dimensional image information to obtain three-dimensional image information, c) evaluating a plurality of temporally successively determined three-dimensional image information for recognizing a gesture.
    [8]
    8. The method according to claim 7, characterized in that the detection of the two-dimensional image information by means of in-infrared cameras takes place.
    [9]
    9. The method according to claim 7, characterized in that the detection of the two-dimensional image information takes place in each case with the aid of a thermopile sensor.
    [10]
    10. The method according to claim 9, characterized in that each thermopile sensor has a resolution of at least 8x8 pixels.
    [11]
    11. The method according to any one of claims 7 to 10, characterized in that when capturing the two-dimensional image information and / or merging image information only within a predetermined temperature range information is taken into account. For this 3 sheets of drawings
    类似技术:
    公开号 | 公开日 | 专利标题
    AT15034U1|2016-11-15|Arrangement and method for recognizing gestures
    DE10361341B4|2007-09-13|Device for operating a hob
    DE102010038341A1|2012-01-26|Video surveillance system and method for configuring a video surveillance system
    DE102013215516B4|2021-11-11|X-ray device and method for controlling an X-ray device
    DE102013012466B4|2019-11-07|Operating system and method for operating a vehicle-side device
    WO2017137104A1|2017-08-17|Home automation system
    EP2952981B1|2019-08-14|Method for operating a building installation with a situation monitor and building installation with a situation monitor
    EP3391163A1|2018-10-24|Device and method for an unmanned flying object
    DE112014007047B4|2020-10-29|Vehicle camera control device
    WO2012095258A1|2012-07-19|Remote-control device and interface module for controlling an apparatus on the basis of a moving object
    DE102017207805A1|2018-11-15|Method for the automated parking of a motor vehicle
    EP2661155B1|2019-07-03|Multifunctional sensor unit and method for the adjustment of said unit
    DE102015012720A1|2017-04-06|Interactive operator system and method for performing an operator action in an interactive operator system
    EP3230828A1|2017-10-18|Detection apparatus for recognizing a gesture and/or a viewing direction of an occupant of a motor vehicle by synchronous driving of lighting units
    EP2898666B1|2017-08-09|Client device for displaying images of a controllable camera, method, computer program and monitoring system comprising said client device
    DE102007041333A1|2009-03-05|Target object i.e. person, contactless counting method for e.g. determination of actual load of production facility, involves identifying breakage or turning of graphical cluster in direction of time axis as object passing area
    EP3584764B1|2021-07-28|Method for controlling a machine by means of at least one spatial coordinate as control variable
    EP2650851A1|2013-10-16|Building automation device
    EP1434184B1|2008-03-05|Control of a multicamera system
    DE102019009080A1|2021-07-01|Setting up sensor locations in camera-based sensors
    WO2017140309A1|2017-08-24|Method for controlling a viewing device arranged on a vehicle in an orientable manner
    DE102020106021A1|2021-09-09|METHOD AND SYSTEM FOR OPERATING A SELECTION MENU OF A GRAPHIC USER INTERFACE BASED ON THE CAPTURE OF A ROTATING CLEAR GESTURE
    DE102014219754A1|2015-04-02|Gesture-based industrial monitoring
    EP3097511A1|2016-11-30|Method for detecting a movement path of at least one moving object within a detection region, method for detecting gestures while using such a detection method, and device for carrying out such a detection method
    DE102019105778A1|2020-09-10|Method for classifying objects within a motor vehicle
    同族专利:
    公开号 | 公开日
    DE102013207664A1|2014-10-30|
    引用文献:
    公开号 | 申请日 | 公开日 | 申请人 | 专利标题
    CN108489049A|2018-03-29|2018-09-04|广东美的制冷设备有限公司|Infrared detection distance method, air conditioner and the computer storage media of air conditioner|
    WO2020135816A1|2018-12-29|2020-07-02|青岛海尔空调器有限总公司|Sensing array-based control method and apparatus, storage medium and computer device|AT505882A1|2007-10-03|2009-04-15|Hierzer Andreas|MOTORIZED LIGHT|DE202018102272U1|2018-04-05|2019-07-10|Sick Ag|Device for monitoring a protective field|
    DE102018109828B4|2018-04-05|2022-02-17|Sick Ag|Device for determining a distance to an object in a room|
    DE102019121863A1|2019-08-14|2021-02-18|Löwen Entertainment GmbH|Lighting device|
    法律状态:
    2019-12-15| MM01| Lapse because of not paying annual fees|Effective date: 20190430 |
    优先权:
    申请号 | 申请日 | 专利标题
    DE102013207664.9A|DE102013207664A1|2013-04-26|2013-04-26|Arrangement and method for recognizing gestures|
    [返回顶部]