WO2020229190A1

WO2020229190A1 - Identifying an object on the basis of recognising part of the object and on the basis of description data of a reference object

Info

Publication number: WO2020229190A1
Application number: PCT/EP2020/061967
Authority: WO
Inventors: Heinz Macher; Urs Hunziker
Original assignee: Bircher Reglomat Ag
Priority date: 2019-05-10
Filing date: 2020-04-29
Publication date: 2020-11-19
Also published as: DE102019112339A1

Abstract

The invention relates to a sensor system (100) and a method for three-dimensional detection of a scene (190) and for recognising objects (195) in the three-dimensionally detected scene. The sensor system comprises: (a) an illumination means (130) for illuminating the scene with illumination light; (b) a measurement device (110), for example a time of flight (TOF) measurement device, for receiving measured light (196) on a light receiver (120), which measured light is illumination light (196) scattered back by an object, and for measuring distances on the basis of a light transit time of the illumination light and measured light; (c) a data-processing device (150) for ascertaining a three-dimensional characteristic of the scene on the basis of the measured distances; and (e) a data memory (160) in which at least one description of at least one reference object is stored. With respect to programming, the data-processing device (150) is designed: (i) to recognise at least one portion of the object (195); (ii) if merely one such portion was recognised, said device is designed to compare the recognised portion with the at least one description; and (iii) if a match is established between the recognised portion and a reference portion of the at least one reference object during the comparing process, said device is designed to identify the object. The invention also relates to the use of the sensor system for controlling a covering characteristic of an opening (184) to be passed by an object, by at least one closing body (186). The closing body is designed for example as a two-part sliding door. During operation of the sensor system, the data-processing device (150) controls two motors (187). The object type to which access is intended to be granted can be a human, for example. Access to or passage through the opening (184) is intended to be prevented for other object types, for example animals.

Description

Identification of an object based on recognition of a part of the object and on description data from a reference object

Technical area

The present invention relates to the technical field of object recognition by means of recording three-dimensional images and subsequent image evaluation.

Background of the invention

To open and / or close openings are often means

Actuators operated closure bodies are used, which facilitate the handling of the relevant closure body for operators or are operated automatically without any operating action, for example when an object to be passed through the opening comes into the area of the opening. Such an opening can be a passage in a building, for example. A closing body can be a door, for example.

As a rule, such a closure body should only be caused to move when a certain object or a certain object is in a certain detection area of a sensor system assigned to the opening

certain type of object is recognized. In some applications, for example, a door should be opened automatically when a person approaches this door; if an animal approaches this door, then this door should remain locked.

AD: PS: vg From EP 2 453 252 B1 a 3D sensor system is known for the application area of monitoring automatically opening doors, which is based on the principle of measuring the transit time of light beams,

Illumination sources are emitted and detected by a light receiver after an at least partial reflection or 180 ° backscatter. Such sensor systems are generally referred to as "time-of-flight" (TOF) sensor systems. However, TOF sensor systems have the disadvantage that with

As the distance d of the object to be detected increases, the intensity of the measurement light to be detected (backscattered) by a light receiver of the TOF sensor is weakened in two ways. In the case of a point illumination light source without special focusing, this is scaled

Attenuation of the illumination light emitted by the illumination sources with l / d ^/ 2, where d is the distance to the illumination light source. The same applies to the measuring light if you have a point on the object where the

Illuminating light is scattered isotropically, perceived as a point light source. As a result, in this case this leads to a l / d 'scaling of the intensity of the

received measurement light.

For energy reasons, lighting sources for sensor systems for monitoring doors can only provide a limited intensity of lighting light. In the case of TOF 3D sensor systems, in particular, this means that object recognition is often not sufficient to distinguish whether a certain type of object is located in the detection area of the sensor system.

The present invention is based on the object in the field of automatic monitoring of openings, which by means of at least one

Closing body can be closed to enable reliable object identification. Summary of the invention

This task is solved by the subjects of the independent

Claims. Advantageous embodiments of the present invention are described in the dependent claims.

According to a first aspect of the invention, a sensor system is described for three-dimensional recording of a scene and for recognizing or

Identifying objects in the three-dimensionally captured scene. The sensor system described has (a) an illumination device for illuminating the scene with illuminating light; (b) a measuring device (bl) for receiving measuring light, which is at least partially back-scattered illumination light from at least one object contained in the scene, and (b2) for measuring distances between the sensor system and the at least one object based on a light transit time of the illumination light and the measuring light; (c) a data processing device connected downstream of the measuring device for determining a three-dimensional

Characteristics of the scene based on the measured distances; and (d) a data memory which is coupled to the data processing device in terms of information technology and in which at least one description of at least one reference object is stored. The data processing device is set up in terms of programming (i) to recognize at least a partial area of the object; (ii) if only such a sub-area was recognized, to compare the recognized sub-area with the at least one description; and (iii) if during the comparison (within a predetermined accuracy) a match is found between the recognized sub-area and a reference sub-area of one of the at least one reference object, for identifying the object as an object corresponding to the corresponding at least one reference object.

The sensor system described is based on the knowledge that a Incomplete (intentional or unintentional) scene detection, in which an object located in the scene is not fully recognized, through a priori knowledge of the appearance of objects that may be in the scene, can be sufficient to carry out an object identification. Expressed clearly, a comparison with can be carried out using an image evaluation following the detection of only a partial area

Reference subareas of one or more reference objects are made. Depending on the result of the comparison, this comparison can lead to an identification of the object located in the scene. This can be used in many applications, even with poor quality scenes

Detections or object recognition a reliable object identification can be made.

In this context, an incomplete scene detection can also be a bad scene detection in at least partial regions with a low one

Accuracy that is insufficient to fully recognize an object located therein. However, this does not necessarily have to be due to the

Scene detection by the measuring device. It is also possible that of one and the same object due to different optical

Properties of various parts of this object are not all

Partial areas of the object can be recognized.

The term “scene” can in particular be understood to mean that spatial area which is evaluated by the sensor system using image processing methods. Objects located in the scene are recognized at least in partial areas by means of a suitable image evaluation. For this purpose, the data processing device can use known methods for

Image evaluation and / or image analysis can be used. The

Data processing device can therefore be a special one

Be an image processing processor or have one that is configured, known methods for image evaluation and / or image processing apply or carry out.

In this document, the term “illuminating light” is to be understood as meaning those electromagnetic waves which are emitted by a light source of the illuminating device and strike the relevant object in the scene. The "measuring light" are the electromagnetic waves scattered back from or on the object, which are emitted by the measuring device or a

Received light receiver of the measuring device and used for the three-dimensional evaluation of the scene.

The terms “optical” and / or “light” can refer to electromagnetic waves that have a specific wavelength or frequency or a specific spectrum of wavelengths or frequencies. In particular, the electromagnetic waves used can be assigned to the spectral range that is visible to the human eye. Alternatively or in combination, electromagnetic waves assigned to the ultraviolet (UV) or the infrared (IR) spectral range can also be used. The IR spectral range can extend into the long-wave IR range with wavelengths between 3.5 pm to 15 pm, which by means of the

Light receiver of the sensor can be detected.

The term “object” can be understood to mean any three-dimensional physical structure which at least partially has a surface quality that leads to at least partial reflection or scattering of illuminating light and is thus visible to the measuring device through the resulting measuring light. The object can be an object such as a

Be a motor vehicle or a living being such as a person. The object can be an object that is static or stationary with respect to the sensor system. Furthermore, the object can also move within the scene, leave it or enter it. A repeated scene capture can then (by comparing the different spatial positions) the movement (according to the law

Speed = distance / time) of the object. Depending on the application, the absolute value of the speed and / or the

Motion vector, i.e. additionally the direction of movement can be determined.

The term "information technology downstream connection" or "information technology coupling" can be understood in this document as an electronic and / or optical connection which enables an exchange of analog or preferably digital data signals between the components involved. By means of such data signals, a component can cause another component to carry out a specific action.

It should be noted that the described object identification can also be carried out if not only one (contiguous) sub-area of the object but two or more (spatially separated sub-areas) are recognized. In this case, the description of the at least one reference object with regard to several

Reference subareas examined. As a result, the reliability of the object identification can be increased when a plurality of object subareas are recognized.

According to an exemplary embodiment of the invention, the data processing device is also set up in terms of programming (iv) to identify the object as an unknown object if, during the comparison (within the specified accuracy), no correspondence is found between the recognized sub-area and a reference sub-area of one of the at least one reference object becomes.

Identification as an unknown object can be made in building security, for example. An unknown object can, for example, be restricted by authorization to pass through an opening a closure body can be closed, be excluded. For this purpose, the closing body can remain in a closed position or be transferred from an open position to a closed position.

According to a further exemplary embodiment of the invention, the at least one description has a geometric description.

The use of geometric descriptions has the advantage that reference objects can each be described with a compact data set that has a comparatively small data size. With a level of reliability that is completely sufficient for most applications, this enables the required image processing operations to be carried out quickly, even with limited computing power of the data processing device, and thus very quick identification. This extends the scope of the

described sensor system, because a very fast object identification is required in many applications.

According to a further embodiment of the invention, the

Data processing device set up in terms of programming to carry out the comparison of the recognized partial area with the at least one description on the basis of three-dimensional data.

A comparison on the basis of two-dimensional data is usually simpler in terms of computation and, in particular, can be carried out more quickly. In many applications, however, a "3D comparison" based on three-dimensional data has proven to be much more accurate, so that the process reliability when identifying the object is correspondingly greater. Of course, in addition to the inventive 3D detection of the scene, a 3D comparison also requires (i) corresponding 3D data processing leading to 3D recognition of at least the sub-area of the object and (ii) a 3D description of the at least one reference object. According to a further exemplary embodiment of the invention, the lighting device has (a) a laser device

Illumination light source for spatial scanning of the scene with a

emitted laser beam illuminating light, (b) an at least approximately point-shaped illuminating light source, (c) a plurality of individual

Illumination light sources, which are in particular individually controllable and each assigned to a specific solid angle range of the scene, and / or (d) a flat illumination light source, in particular with a light intensity that is not homogeneous over the area.

A laser beam that scans the scene can be directed in a known manner via two rotatable mirrors with axes of rotation that are not parallel to one another and preferably oriented perpendicular to one another to the point of the scene to be illuminated. For such a (dynamically adaptive) deflection, non-mechanical optical elements such as diffractives can also be used

Optical elements (DOEs) are used. The deflection can in particular be controlled by the above-described illuminating light control device.

The at least approximately point-shaped illumination light source can be a (sufficiently strong) semiconductor diode, for example a laser or light-emitting diode. In order to specifically illuminate the scene over a large area, suitable

Beam shaping systems, in particular lens systems, are used. In order to achieve the uneven lighting of the scene depending on the spatial angle, suitable optical elements for beam deflection,

Beam splitting and / or beam merging can be used. DOEs can also be used to advantage.

The plurality of illumination light sources, which are likewise in particular lasers or light-emitting diodes, can be controlled (in particular individually) by the illumination light control device described above. This allowed in an advantageous manner, by means of one described below

Illumination light control device, an adaptively controlled or even regulated setting of a characteristic of the illumination light.

A flat light source can also be the source for an intensity distribution that is not homogeneous as a function of the spatial angle. If it is a spatially homogeneously illuminated surface, suitable optical elements can be used

Beam deflection, beam splitting, beam merging and / or beam shaping can be used in order to achieve the described non-uniform illumination of the scene as a function of the spatial angle.

According to a further embodiment of the invention, the

Lighting device can be operated (i) in a first lighting operating mode for illuminating the scene with a first illumination and (ii) in a second lighting operating mode for illuminating the scene with a second illumination. The detection reliability for the at least one partial area of the object in the second illumination is greater than that

Detection reliability with the first illumination. This has the advantage that the scene can then be better illuminated adaptively if (in the future or shortly) object recognition is to be expected. A prognosis for an object detection can be based on a knowledge or a guess about the

Presence of objects based. Alternatively or in combination, the prognosis can also be based on a detection of a partial area that has just taken place. The operation of the

Lighting device in the second lighting operating mode can then, in many application cases, result in a larger partial area of the object (or even the entire object) being able to be recognized during the next detection with better illumination. This increases the reliability of the object identification.

The illumination of the scene can affect the intensity of the illuminating light refer, which is provided by the entire lighting system. The illumination of the scene can, for example, by the

Use of an adaptive optical element, in certain regions of the scene where an object is suspected to be larger than in other regions. Furthermore, the illumination can be changed by changing a repetition rate of a plurality of successive exposures. Such an exposure can have an areal illumination of the scene or a sequential scanning of the scene with a laser beam deflected, for example, by two mirrors.

It should be noted that, as a rule, the energy requirements of the

Lighting device and therefore the energy consumption of the whole

Sensor system are larger when the second lighting operating mode is activated. Since this will often only be the case in relatively short time windows over a longer period of time, the average energy consumption of the

Sensor system can be kept low. In other words, the described sensor system can be operated most of the time with the first lighting operating mode, so that in comparison to a continuously

existing strong second illumination the energy consumption is significantly lower.

According to a further embodiment of the invention, the

Sensor system further to an illumination light control device which is information-technically coupled to the data processing device, the data processing device also being set up in terms of program technology to cause the illumination light control device to switch from the first illumination operating mode to the second illumination operating mode when a partial area of the object is detected .

Is clearly expressed in this embodiment after a

Object recognition produced an improved lighting situation, so that with a subsequent object recognition, if necessary, a larger sub-area or further sub-areas of the object can be recognized. Such a spatially extended detection can then improve the

Reliability of the identification of the object.

According to a further embodiment of the invention, the

Measuring device on a light receiver, which can be operated (i) in a first light receiving operating mode for receiving measuring light with a first sensitivity and (ii) in a second light receiving operating mode for receiving measuring light with a second sensitivity, the detection reliability for the at least a partial area of the object is larger at the second sensitivity than at the first sensitivity. This has the advantage that, in analogy to the above-described operating modes with regard to the lighting, the scene can then be adaptively recorded more precisely when object recognition is to be expected (in the future or shortly). In this case too, a prognosis for an object recognition can be based on knowledge or a presumption about the presence of objects and / or on a recognition of a sub-area that has just taken place. The operation of the light receiver in the second light receiving operating mode caused by such a prognosis can then contribute to the fact that a larger sub-area of the object (or even the entire object) can be recognized during the next detection. This increases the reliability of the object identification (further).

According to a further embodiment of the invention, the

Sensor system furthermore to a light receiver control device which is coupled to the data processing device and to the light receiver. In terms of programming, the data processing device is also set up to cause the light receiver control device to switch from the first light reception operating mode to the second light reception operating mode upon detection of a partial area of the object. In this exemplary embodiment, in analogy to the switchover of the lighting operating mode described above after object recognition, the operation of the light receiver is clearly expressed to the second

Light reception operating mode changed over, so that with a subsequent object detection, a larger sub-area or more, if necessary

Partial areas of the object can be recognized. This can also bring about an improvement in the reliability of the identification of the object.

According to a further embodiment of the invention, the

The light receiver has a plurality of pixels for receiving the measurement light, wherein in one of the two light reception operating modes at least two pixels of the plurality of pixels are combined to form a superordinate pixel.

Typically, at least some of the plurality of pixels are combined in such a way that in each case a certain number of pixels are combined to form a superordinate pixel. The certain number can be, for example, (preferably) two, three, (preferably) four, six, (preferably) eight, or (preferably) nine. Of course, an even stronger grouping of pixels is also possible.

Such a grouping of pixels, which is also referred to as "binning", has the effect that, at the expense of spatial resolution, the number of photons of the measurement light that are collected or accumulated by a pixel during a scene detection increases according to the number is increased to a parent pixel combined pixels. This reduces the so-called statistical photon noise, especially in the case of weak measurement light, which improves the scene evaluation accuracy. "Binning" is therefore particularly advantageous in the case of weak measuring light when a high spatial resolution is not required. It is pointed out that binning can also be carried out locally over the area of the light receiver in only at least a partial area of the active areas of the light receiver. Such a local "binning" then leads to an inhomogeneous spatial resolution, which is not absolutely desirable. The disadvantage of such an inhomogeneous spatial resolution is increased in many applications

Photon accumulation overcompensated. A local “binning” can, at least in the case of some known light receivers, without special electronic or apparatus elements, simply by a corresponding activation of the light receiver by the light receiver control device, if necessary

Initiated by the data processing device, which control determines the "binning" and thus the light reception operating mode.

In preferred embodiments, local "binning" is carried out to the effect that, measured by the measuring device and / or learned by the data processing device, precisely those areas of the

Light receivers, which have received too little light energy in at least one previous scene detection, by suitable control of the light receiver by the light receiver control device in subsequent scene detections in a suitable manner to superordinate pixels

be summarized. Such a dynamically controlled or regulated "binning" can be (learned) during normal operation of the sensor system and / or during the configuration of the sensor system, for example in the context of (initial) installation, maintenance, cyclical or

automatic reconfiguration etc. can be carried out.

It is also pointed out that with a non-square number of individual pixels combined to form a superordinate pixel, the spatial resolution of the light receiver is different along different directions if the individual pixels have a square shape. In some applications, this can be advantageous be exploited. Such an application is, for example, when a movement of an object in the scene along a previously known spatial direction is to be detected with high accuracy than a movement along another spatial direction, preferably perpendicular thereto. In such a case, the number of pixels which are arranged along a line parallel to this previously known spatial direction (as it is depicted on the light receiver) can be greater than the number of pixels which are arranged along a line perpendicular thereto. Then the spatial resolution along the direction of movement is greater than the spatial resolution perpendicular to the direction of movement and the movement profile of such a linearly moving object can be determined with a particularly high degree of accuracy even with a comparatively weak measuring light.

It is also pointed out that the described binning can also be adapted adaptively in response to at least one previously recorded (and evaluated)

Scene characteristics can be activated (automatically). This means that the "binning" is not only controlled by the light receiver control device but is regulated as a function of the results obtained through a scene evaluation. This makes a particularly reliable one

Enables scene detection even with weak measurement light, so that the sensor system described can also be used with a correspondingly weak one

Illuminating light and thus can be operated in an energy-efficient manner.

According to a further embodiment of the invention, the

Data processing device further configured such that a

Covering characteristic of an opening can be controlled by at least one closure body. As a result, the opening, which is, for example, an entrance (or an exit) of a building, can be automatically monitored in a particularly reliable manner. A suitable control of a

Actuator, the closing body can be moved automatically between an open position and a closed position. The Data processing device of the sensor system described are coupled to the control of a known control system for a closure body.

The wear body can be any desired barrier element by means of which an opening can be temporarily at least partially closed. The closing body can be, for example, (i) a sliding door, (ii) a revolving door, (iii) a turnstile, (iv) a gate, in particular a garage door, (v) a barrier, in particular at a level crossing or a vehicle entrance or

Vehicle exit.

According to a further aspect of the invention, a method is described for three-dimensional detection of a scene and for recognizing objects in the three-dimensionally detected scene by means of a sensor system and in particular by means of a sensor system of the type described above. The

The method described comprises (a) illuminating the scene with

Illuminating light using a lighting device; (b) receiving measurement light, which is at least partially backscattered illumination light from at least one object contained in the scene, using a measurement device; (c) a measurement of distances between the sensor system and the at least one object based on a light transit time of the illuminating light and the measuring light, using the

Measuring device; (d) determining a three-dimensional characteristic of the scene based on the measured distances using a

Data processing device which is connected downstream of the measuring device for information purposes; (e) recognition of at least a partial area of the object using the data processing device, which

is information-technically coupled to a data memory in which at least one description of at least one reference object is stored; (f) if only such a sub-area was recognized, a comparison of the recognized sub-area with the at least one description; and (g) if at the comparison (within a predetermined accuracy) a match between the recognized sub-area and a

Reference sub-area of one of the at least one reference object was determined, identifying the object as an object corresponding to the corresponding at least one reference object.

The described method is also based on the knowledge that incomplete or imprecise scene detection can be sufficient to carry out an object identification based on recognized object subareas taking into account a priori knowledge stored in the data memory about possible appearance forms of the object.

According to an exemplary embodiment of the invention, the object is an object moving in the scene. This means that the described sensor system is used to detect a scene whose characteristics change over time. With such a time-varying scene characteristic, the advantages of the sensor system described here come into play to a particular degree. A scene with a characteristic that changes over time can only be sensed in a meaningful manner if a time sequence of a plurality of scene acquisitions is carried out from which the movement of the object can be derived. Since, especially with fast moving objects, the

If illuminating light is required, the measuring light which is collected in the light receiver during an individual measurement is correspondingly weak. This can help to ensure that not the entire object but only a part of it is recognized. The a priori knowledge about possible object manifestations stored in the data memory is then used according to the invention to recognize or identify the entire object.

According to a further exemplary embodiment of the invention, the object is a Living being, especially a human.

When identifying a person who, for whatever reason, cannot or has not been fully recorded, the method described here can be used in a particularly advantageous manner. A person has several limbs that have very different dimensions with regard to their size. A living being and

a person in particular can be described physiologically very precisely. Corresponding description data are stored in the data memory. For example, it can be taken into account that fingers with a minimum detection volume of typically a few centiliters are naturally always connected to arms less than 1 meter in length. These then lead to a body that has a minimum detection volume of several liters. Therefore, with the method described here, protection against a hazard from a moving closure body can be implemented in a simple manner, which moves in particular in the direction of a closed position in order to close an opening, for example a passage opening in a building. In this case it can be concluded from a detection of several liters of "abnormal volume" (= volume that is present in addition to the "empty volume" without the object) that the fingers of a human person must be within a radius of one meter . To be in a reliable manner

To prevent pinching of fingers must be done in a case where the said "abnormal volume" is at a distance of less than 1 meter from the

Closing body is recognized, a movement of the closing body can be prevented.

The principle described above using the example of finger and body can of course be expanded to include the body part "arm": Volume differences of 1: 10: 100 are typically present for the volumetric detection of finger - arm - body. A corresponding three-stage object identification can however, implement themselves with the same principle. This also applies to other combinations of body parts, for example a head, a leg, a foot, etc.

The above-described “model transition” in the case of “figurative” object recognition can also increase the security of the reliable recognition when interpreting a point cloud from 3D point data. This is important to the extent that such a point cloud is an intermediate data format that is standard in many applications and is used in a 3D scene analysis.

According to a further aspect of the invention, a use of a sensor system of the type described above for controlling a covering characteristic of an opening to be passed by an object by at least one closure body is described.

The use described is based on the knowledge that a

Object identification on the basis of partial object recognition and an a priori knowledge stored in the data memory in an advantageous manner

Passages can be used which can be closed by a closure body. This applies in particular to passages which have a sealing or covering characteristic that is different from the

described sensor system is controlled or at least co-controlled.

According to an exemplary embodiment of the invention, the opening is an entrance or an exit, in particular an emergency exit in a building.

By recognizing an existing but possibly not moving object in a passage area, an entrance or exit

monitored, in particular a blocked emergency exit detected, and the

corresponding information to an affiliated system, for example to a Monitoring system.

According to a further exemplary embodiment of the invention, the object is a person or a vehicle. In this case, the building can in particular be a house or a garage.

Further advantages and features of the present invention emerge from the following exemplary description of currently preferred embodiments.

Brief description of the drawing

Figure 1 shows the use of a sensor system for controlling a

Covering characteristic of an opening by means of closing bodies designed as sliding doors.

FIG. 2 shows a person with several recognizable partial areas which can be used to identify the object “person”.

Figures 3a and 3b illustrate a combination of individual pixels for

Increase in the light sensitivity of a light receiver.

Detailed description

FIG. 1 shows the use of a sensor system 100 for controlling a coverage characteristic of an opening 184 by a closure body 186 which is designed as a two-part sliding door. According to the exemplary embodiment shown here, the opening 184 is an entry opening for people into a building or a garage entrance for motor vehicles. The corresponding input structure is provided with the reference number 180. An in an object 195 located by the sensor system 100 monitored scene 190 is intended to symbolize such a person or a motor vehicle.

The entrance structure 180 comprises a stationary support structure 182 which has a frame and a guide for the two sliding doors of the

Closing body 186 represents. The sliding doors 186 can each be opened by means of a motor 187 along the lines shown by two thick double arrows

Displacement directions are moved. The motors 187 are controlled, as explained below, by means of a data processing device 150 of the sensor system 100 described in this document. The data

Processing device 150 thus also serves as a control device for closing body 186.

The sensor system 100 has a time of flight (TOF) measuring device 110, the data processing device 150 and a data memory 160.

The TOF measuring device 110 in turn has an illumination device 130 and a light receiver 120. According to the one shown here

The TOF measuring device 110 has or are assigned to the TOF measuring device 110 (i) an illumination light control device 135 for controlling the operation of the illumination device 130, (ii) a measuring unit 125 connected downstream of the light receiver 120 for measuring a

Time of flight between the illumination light 131 emitted by the illumination device 130 and the measurement light 196 received by the light receiver 120 after scattering on the object 195 and (iii) a light receiver control device 140 for controlling the operation or for selecting an operating mode of the light receiver 120.

In the TOF measuring device 110, all optical components of the

Sensor system 100 housed. The entire sensor system 100 (in contrast to the illustration in FIG. 1) is preferably constructed as a module which also has the data processing device 150 and the data memory 160 within a compact design in addition to the TOF measuring device 110.

When the sensor system is in operation, the data processing device 150 controls the two motors 187. The electrical power required for operating the doors 186 is provided by an output stage or a power amplifier which, according to the exemplary embodiment shown here, is in the housing of the respective motor 187 is integrated. The two output stages are each controlled via a control signal 152a, which is sent to one

Control output 152 is output from the data processing device 150.

According to the exemplary embodiment shown here, the sensor system 100 is able to carry out an object detection. The data is used for this

processing device 150 to a data record of reference objects stored in the data memory 160, which correspond to selected objects that are authorized to pass through the opening 184. This means that when the object 195 appropriately approaches the entrance 184, the sliding doors 186 are only opened when the recognized object 195 at least approximately matches one of the stored reference objects. This clearly means that when using the

Sensor system 100 an object-based access control takes place.

Consistent with the spirit of this document

Descriptions of the reference objects are stored in the data memory 160. Furthermore, the data processing device 150 is set up in terms of program technology, if only (at least) a partial area of the object 195 was recognized, to compare this partial area with the descriptions. The recognized sub-area is characterized with known means of image analysis and one with different parts Description compared. For this purpose, each description of a reference object can comprise a plurality of partial descriptions, the partial descriptions also having a certain amount with regard to the structure of the respective reference object

May have overlap. This has the advantage that a large number of partial descriptions can be assigned to the reference object, each of which can be individually compared with the characteristic of the recognized partial area. When detected within a certain accuracy

If the recognized partial area corresponds to a partial description of a reference object, the object 195 located in the scene 190 is identified as an object which corresponds to the reference object at least in terms of its type.

It should be noted that in the data memory 160, if a

Access control should only take place in relation to one type of object 195, and only a description of a single reference object can be stored (possibly with a plurality of partial descriptions). This means that the sensor system 100 monitors the access for an object type based on at least one detected object sub-area. The type of object to which access is to be made possible can, for example, be a person. Access to or passage through the opening 184 is to be denied for other types of objects, such as animals.

FIG. 2 shows an object type human 195 with several recognizable ones

Sub-areas 295a, 295b, 295c, 295d and 295e. According to the exemplary embodiment shown here, the partial areas are assigned to characteristic structures of the human body 195, which are particularly simple and

can be reliably recognized. For reasons of clarity, not all (possible) partial areas that can be used for object identification on the basis of recognized partial areas are shown. In particular, in addition to the partial area head 295a, not only the partial areas of the left elbow 295b and left hand 295c but also the partial areas of the right Elbow and right hand are used. The same naturally also applies to the partial areas of the left knee and left foot not shown, which anatomically correspond to the right knee 295d or the right foot 295e.

The trunk of the person 195 is a further sub-area that could be used for object identification. This one has

comparatively few characteristic structures, but can be recognized by the size of its volume.

In order to be able to identify different partial areas 295a, 295b, 295c, 295d and 295e of the object 195 as well as possible, the one shown in FIG

Light receivers are operated in different light receiving operating modes for receiving measuring light. In a first light reception operating mode with a first sensitivity, the scene can be captured particularly quickly. The accuracy of the scene detection is reduced, however, so that there is only limited detection reliability for a partial area of the object 195. In a second light reception operating mode, the scene 190 is recorded with increased accuracy. The detection reliability is improved accordingly. In this context, the data processing device 150 is able to control the light receiver control device 125 in the event of a (not particularly precise) detection of a partial area 295a-e of the object 195

cause to switch from the first light receiving operating mode to the second light receiving operating mode. Then at least the

relevant sub-area recorded again with higher accuracy. The corresponding more precise data are then used for a comparison with the description of a reference object.

According to the exemplary embodiment shown here, the accuracy of the scene detection is determined by the spatial resolution of the light receiver 120. This is namely able to combine neighboring pixels in the first light reception operating mode, so that per combined pixel with a certain intensity of measuring light per unit of time more photons are accumulated.

Figures 3a and 3b illustrate such a combination of

Individual pixels of a semiconductor or CCD chip

Light receiver 320a or 320b. The light receiver 320a has a multiplicity of light-sensitive or photon-collecting pixels 322a. According to the exemplary embodiment shown here, the pixels 322a are assigned to a full spatial resolution of the light receiver 320a, which resolution is predetermined by the semiconductor architecture of the chip 320a.

In the light receiver 320b, four of the light-sensitive pixels (for a full resolution) are associated with a higher-level pixel 322b (for an increased

Photon accumulation per pixel at the expense of a reduced spatial

Resolution). In illustrative terms, a pixel 322b collects four times the amount of light compared to a single pixel 322a. Such a combination (English "binning") reduces the required

(Minimum) intensity of the detected measuring light which is required for evaluating the corresponding image area of the scene 190. Since the intensity of the measurement light 196 depends directly on the intensity of the illuminating light 131, the intensity of the illuminating light 131 can be reduced in the first light receiving operating mode by means of the "binning" and thus the energy consumption of the sensor system 100 can be reduced.

It should be noted that more than two different

Light reception operating modes, each with a different strength

Summary of pixels can be used. It is also possible to combine a different number of individual pixels to form a superordinate pixel in different partial areas of the light receiver. Then individual parts of the scene with a higher spatial resolution (and a lower photon accumulation) and other parts of the scene with a lower spatial resolution (and a higher photon accumulation) are recorded. The described local and differently strong grouping of pixels can also be carried out dynamically or adaptively in precisely those partial areas in which a certain object is currently located.

It is noted that the term “having” does not exclude other elements and that the “a” does not exclude a plurality. Elements that are described in connection with different exemplary embodiments can also be combined. It should also be noted that any reference signs in the claims should not be construed as limiting the scope of the claims.

REFERENCE MARK:

100 sensor system

110 TOF measuring device

120 light receivers

125 measuring unit

130 lighting device

131 illumination light

135 Illumination light controller

140 Light Receiver Control Device

150 data processing device

152 control output

152a control signal

160 data storage

180 entrance structure

182 stationary support structure

184 opening

186 locking body / sliding door

187 engine M

190 scene

195 object

196 measuring light

295a-e sub-areas

320a / b light receiver / sensor chip

322a pixels

322b parent pixel / pooled pixel

Claims

Patent claims

1. Sensor system (100) for three-dimensional recording of a scene (190) and for recognizing objects (195) in the three-dimensionally recorded scene (190), comprising the sensor system (100)

an illumination device (130) for illuminating the scene (190) with illuminating light (131);

a measuring device (110)

for receiving measurement light (196) which is at least partially backscattered from at least one object (195) contained in the scene (190)

Illuminating light (196) is, and

for measuring distances between the sensor system (100) and the at least one object (195) based on a light transit time of the

Illuminating light (131) and the measuring light (195);

a data processing device (150) connected downstream of the measuring device (110) in terms of information technology for determining a three-dimensional characteristic of the scene (190) based on the measured distances; and a data memory (160), which in terms of information technology with the

Data processing device (150) is coupled and in which at least one description of at least one reference object is stored; wherein the data processing device (150) is set up in terms of programming

(i) for recognizing at least a partial area (295a-e) of the object (195);

(ii) if only such a sub-area (295a-e) was recognized, for

Comparing the recognized partial area (295a-e) with the at least one description; and

(iii) if, during the comparison, a match is found between the recognized sub-area (295a-e) and a reference sub-area of one of the at least one reference object, to identify the object (195) as an object (195) corresponding to the corresponding at least one reference object.

2. Sensor system (100) according to the preceding claim, wherein the data processing device (150) is also set up in terms of programming

(iv) if no correspondence is found between the recognized partial area (295a-295e) and a reference partial area of one of the at least one reference object during the comparison, for identifying the object (195) as an unknown object.

3. Sensor system (100) according to one of the preceding claims, wherein the at least one description has a geometric description.

4. Sensor system (100) according to one of the preceding claims, wherein the data processing device (150) is set up in terms of programming to carry out the comparison of the recognized partial area (295a-e) with the at least one description on the basis of three-dimensional data.

5. Sensor system (100) according to one of the preceding claims, wherein the lighting device (130)

(a) an illumination light source designed as a laser for spatially scanning the scene with an emitted laser beam illumination light (131),

(b) an at least approximately point-shaped illuminating light source,

(c) a plurality of individual illumination light sources, which in particular can be individually controlled and are each assigned to a specific solid angle range of the scene, and / or

(d) a planar illumination light source, in particular with a light intensity that is not homogeneous over the area.

6. Sensor system (100) according to one of the preceding claims, wherein the lighting device (130) can be operated

(i) in a first lighting operating mode for lighting the scene (190) with a first lighting and (ii) in a second lighting operating mode for lighting the scene (190) with a second lighting, wherein

the detection reliability for the at least one partial area (295a-295e) of the object (195) in the second illumination is greater than in the first

Illumination.

7. The sensor system (100) according to the preceding claim, further comprising an illumination light control device (135) which is information technology-related to the data processing device (150), wherein

the data processing device (150) is also set up in terms of program technology to cause the lighting control device (135) to switch from the first lighting operating mode to the second lighting operating mode upon detection of a partial area (295a-295e) of the object (195).

8. Sensor system (100) according to one of the preceding claims, wherein the measuring device (110)

a light receiver (120) which is operable

(i) in a first light receiving operating mode for receiving measuring light (196) with a first sensitivity and

(ii) in a second light receiving operating mode for receiving measuring light (196) with a second sensitivity, the detection reliability for the at least one sub-area (295a-295e) of the object (195) being greater for the second sensitivity than for the first sensitivity .

9. Sensor system (100) according to the preceding claim, further comprising a light receiver control device (140), which with the

Data processing device and is coupled to the light receiver (120), wherein

the data processing device (150) is also set up in terms of program technology, when a partial area (295a-295e) of the object (195) is recognized, the To cause light receiver controller (140) from the first

To switch light receiving operating mode to the second light receiving operating mode.

10. Sensor system (100) according to one of the two preceding claims, wherein

the light receiver (120; 320a, 320b) has a plurality of pixels (322a) for receiving the measurement light (196), wherein

in one of the two light reception operating modes, at least two pixels (322a) of the plurality of pixels (322a) form a superordinate pixel (322b)

are summarized.

11. The sensor system (100) according to any one of the preceding claims, wherein the data processing device (150) is further configured such that a coverage characteristic of an opening (184) by at least one

Closing body (186) is controllable.

12. Method for three-dimensional detection of a scene (190) and for

Detecting objects (195) in the three-dimensionally captured scene by means of a sensor system (100), in particular a sensor system according to one of the preceding claims, having the method

Illuminate the scene (190) with illuminating light (131) below

Using a lighting device (130);

Receiving measurement light (196), which is at least partially backscattered illumination light (196) from at least one object (195) contained in the scene (190), using a measuring device (110);

Measuring distances between the sensor system (100) and the at least one object (195) based on a light transit time of the

Illuminating light (131) and the measuring light (195) using the measuring device (110);

Determination of a three-dimensional characteristic of the scene (190) based on the measured distances using a data processing device (150) which is connected downstream of the measuring device (110) in terms of information technology;

Recognition of at least a partial area (295a-e) of the object (195) using the data processing device (150), which

is information technology coupled to a data memory (160) in which at least one description of at least one reference object is stored; if only one such sub-area (295a-e) was recognized, comparing the recognized sub-area (295a-e) with the at least one description; and

if, during the comparison, a match between the recognized sub-area (295a-e) and a reference sub-area of one of the at least one reference object was determined, identifying the object (195) as an object (195) corresponding to the corresponding at least one reference object.

13. The method according to the preceding claim, wherein

the object (195) is a moving object in the scene.

14. The method according to any one of the two preceding claims, wherein the object (195) is a living being, in particular a person.

15. Use of a sensor system (100) according to one of the preceding claims 1 to 11 for controlling a coverage characteristic of an opening (184) to be passed by an object (195) by at least one closure body (186).

16. Use according to the preceding claim, wherein

the opening (184) is an entrance or an exit, in particular a

Emergency exit in a building.

17. Use according to one of the two preceding claims, wherein the object (195) is a person or a vehicle.