DE102007041333A1 - 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 - Google Patents

Abstract

The method involves providing a reference area (A) e.g. frame of a door, with high contrast characteristics against an object to be counted. The area is illuminated by a light source (1), and photos of the area are developed by a camera (2) and an evaluating device (3) at a point of time. The photos are merged along a time axis to a flow pattern, and the images of the objects are connected with each other to form a graphical cluster. Breakage or turning of the cluster in the direction of the time axis of the pattern is identified as the object passing the area. An independent claim is also included for a device for contactless counting of a target object i.e. person.

Description

Technical area

The The invention relates to a method for contactless counting of Zählobjekten, preferably by persons by means of a camera-based evaluation device according to the generic term of claim 1 and to a device for carrying out this Method according to the preamble of claim 5.

State of the art

Non-contact counting procedures is of great importance in the determination of current utilization of equipment, premises or campuses as well as for statistical data collection. For example, at exact knowledge of Frequentierhäufigkeit on traffic routes or at the entrances appropriate stadia-related logistical measures are initiated, a crowd or goods rush to manage better.

The currently on the market counting systems in many ways do not meet the requirements of one efficient counting of toughening objects, In particular, results in a high error rate close to each other crowded People and a high computational burden in camera-based systems.

Currently used counting systems deliver, due to shadow throws, optical reflections and other interferences often inaccurate Results. Because the contrast of the camera to be detected counting objects across from whose environment is often deficient, escapes the counting system usually at least part of the passing toughening objects.

According to the state There are different principles of detection in the art the toughening objects, where it is with the toughening objects both persons and animals or goods or vehicles can.

Next a 2D recognition of toughening objects In the range of visible and infrared light can also be a 3D detection using a stereo camera pair and a detection of Light transit times take place.

The Images generated by the camera systems are controlled by a processor Analyzing analyzed, with z. B. image region based Statistical method or a quantity analysis of moving pixels or Edges or so-called Tripwire method or a virtual cutting a counting line through a dough object trajectory Find ("tracking") use, around respective toughening objects to recognize.

Besides counting systems exist, which work with pressure-sensitive sensors and light barriers.

Presentation of the invention

It is an object of the present invention, an alternative non-contact counting system to propose, which has a high counting accuracy and especially insensitive to the appearance of shadows and the brightness or darkness of in a respective Observanzbereich located toughening objects or persons.

Of the relatively high technical complexity of known imaging counting systems should be reduced and enabled an effective evaluation of image data.

According to the invention these objectives by a method having the features of the claim 1 and by a device for carrying out this method with reached the features of claim 5.

• – mindestens ein Referenzbereich vorgesehen ist, welcher gegenüber einem zu erfassenden Zählobjekt höhere Kontrasteigenschaften aufweist als ein zum Referenzbereich benachbarter Umgebungsbereich
• – wobei der Referenzbereich von einer Lichtquelle beleuchtet wird
• – und eine Kamera sowie eine daran assoziierte Auswerteeinrichtung vorgesehen sind, welche zu aufeinander folgenden Zeitpunkten Bilder des Referenzbereichs erstellen
• – und die Bilder entlang einer Zeitachse zu einem Flussbild zusammengefügt werden
• – wodurch Abbildungen eines auf den Bildern aufscheinenden Zählobjektes entlang der Zeitachse jeweils zu einem auf dem Flussbild dargestellten graphischen Cluster miteinander verbunden werden
• – und wobei ein Abreißen oder ein Auftauchen eines graphischen Clusters in Richtung der Zeitachse des Flussbildes als ein Passieren des Referenzbereichs durch ein Zählobjekt bzw. als zu zählendes Ereignis erkannt wird.

A method for the contactless counting of toughening objects, preferably of persons by means of a camera-based evaluation device according to the invention is characterized in that

• - At least one reference area is provided, which has higher contrast properties compared to a to-be-measured object to be detected as an adjacent reference area to the surrounding area
• - Wherein the reference area is illuminated by a light source
• - And a camera and an evaluation device associated therewith are provided, which create at successive times images of the reference area
• - And the images along a time axis are joined together to form a flow chart
• Whereby images of a toughening object appearing on the images are connected to each other along the time axis to form a graphical cluster represented on the flowchart
• And wherein a tearing off or emergence of a graphical cluster in the direction of the time axis of the flow image is detected as a passing of the reference region through a to-be-counted object or as an event to be counted.

Thus, by means of an evaluation device in a mosaic-like manner, a cluster composed sequentially from a plurality of count object images is generated, and then an overflow If a cluster which has appeared on the flow chart breaks off or is not continued by any further cluster segment mappings, it can be determined with little computation effort and in an unambiguous manner whether a toughened object has left this reference area. In this case, a complete passing of the reference range is indicated and a corresponding counting pulse is triggered and forwarded to a respectively designated instance.

Instead of a ripping off of a graphical cluster, it is also possible the emergence of a graphical Cluster or the formation of two or more adjacent ones Illustrations of a toughening object as to be counted by the evaluation device event.

The Compared to the surrounding area increased contrast properties of the reference range are preferably realized by the reference range increased Has reflective properties.

alternative it is also possible a reference area with opposite a to-be-measured object to be detected increased Absorption properties as it has the surrounding area, provided. This variant proves to be in those applications as advantageous in which particularly bright or reflective toughening objects in the respective Observanzbereich are to be expected.

by virtue of the heightened Contrast properties of the reference area compared to its surrounding area results in a high toughness contrast on the pictures taken by the camera. In particular, can the counting system no more of shadows and different brightnesses of the objects to be irritated. An inventive counting system is thus able to ensure a particularly high counting accuracy.

In a development of the counting device according to the invention can also both a reference region with increased reflection properties, as well as a reference area with increased absorption properties, be arranged in combination one behind the other. Since by means of a In such an arrangement, persons dressed especially dark as well as Also particularly brightly dressed persons safely from the evaluation unit can be detected results in an even higher Counting Accuracy the counting method according to the invention.

In a preferred embodiment of the inventive method it is provided that the evaluation device, the flow direction of the graphic cluster and thereby the passing of the the reference area traversing toughening objects recognizes. By observing a single reference range can thus z. B. the number of people entering a room and the number of these premises leaving persons are determined.

In Another preferred variant of the method is provided that the camera at least a portion of the reference range adjacent environmental area in the images, where the position of boundary edges appearing on these images the reference range of the evaluation uniquely located becomes. In this way it is ensured that the boundary edges of the reference range on the images are clearly recorded and thus the reference range for the further image processing is clearly defined.

In a further preferred embodiment the method according to the invention it is intended that the on the camera taken Images of images appearing from the evaluation device due spatial and / or temporal affinity as belonging together detected and a respective toughening object or a (parent) cluster, where those mappings this toughening object which an interruption to a contiguous portion of the associated one (Parent) cluster, either with this (parent) cluster merged in a virtual way or filtered out of the flowchart.

These measure can in the sense of a high counting accuracy be required, depending on the width of the reference area, the outer dimensions of the counting objects, and the picture-taking frequency the circumstance that some of the images or of Toad objects pictures only a small section of the straight entering the reference area Zählobjektes, z. B. a toe show a person. In a mere juxtaposition of such Pictures or pictures would sometimes irritating distances or wave effects between related segment maps, which, however, eliminated by the inventive melting or filtering become.

Around a simple monitoring each observance area and the least possible computational effort to enable is the reference range in a preferred embodiment the invention strip-shaped executed.

An optimized detection of the toughening objects is further ensured by the Refe renzbereich on the ground and the light source above the floor or above the counting to be counted objects is arranged.

By doing as a reference area already existing in the Observatory area Object with opposite the environment increased Reflection or absorption properties is used, results a simplified construction of the counting device according to the invention.

Brief description of the drawings

The Invention will now be explained in more detail with reference to an embodiment. there shows:

1 a schematic representation of a counter device according to the invention with a strip-shaped reference area

2 several to a flowchart according to the invention sequentially assembled camera images

3 a flow chart including two clusters generated according to the invention

4 an alternative arrangement variation of reference ranges according to the invention

Embodiment of the invention

In 1 a device according to the invention for carrying out a contactless counting method is shown schematically.

Here is one in an observance area of the camera 2 , z. B. arranged on the ground reference area A, which has highly reflective properties and z. B. in white color or from an EU standard-compliant covering material for road guidance can be executed. In a preferred embodiment, the reference area A is designed as a self-adhesive element with aluminum coating.

In any case, the reference range has a higher contrast characteristic compared with A to a to-be-measured object than an adjacent to the reference range A surrounding area 4 ,

The reference area A is strip-shaped in the present exemplary embodiment and has a longitudinal extent x of, for example, 2 m. Here it is important that an in 1 In any case, the width y of the reference area A (measured along an expected passing direction) is smaller than the outer dimensions of an expected toughened object or smaller than an average step size of a person. According to the present embodiment, z. B. for counting people a width y of the reference range A of about 30 cm proposed.

One such strip-shaped Reference area A may be approximately across a passageway z. B. lead to a platform or to a function room or premises, so any person crossing the passageway or any toughened object must pass through the reference area A or pass.

Above the bottom or the reference area A are one or more light sources 1 arranged to illuminate the reference area A.

Next to the light source 1 is a camera oriented to the reference area A. 2 arranged, which with a processor-controlled evaluation 3 wired or by radio in connection. At the camera 2 it can be about a commercial CCD or CMOS camera.

The active camera 2 created from its bird's eye view at brief successive times t _{1 ... n} continuously images B _{A1 ... n of} the reference area A and transmits them to the evaluation 3 ,

First, by means of a suitable filtering method, the surrounding area 4 of the reference area A is eliminated from the pictures B _{A1... n} , so that only the area of the reference area A can be seen on each of the pictures B _A1 . For this, the camera can be calibrated to the reference range A before being put into operation. In a preferred variant of the method, it is provided that the camera 2 a partial area of the surrounding area surrounding the reference area A. 4 on the pictures B _{A1 ... n} . In this way, it is ensured that the boundary edges of the reference area A are clearly detected on the images B _{A1... N} , and thus the reference area A is uniquely located for further image processing.

The from the camera 2 created and possibly software-formatted or cropped images B _{A1 ... n} be of the evaluation 3 along a time axis t to a flow chart B joined together, ie juxtaposed or partially superimposed. 2 Fig. 12 schematically shows four images B _A1 , B _A2 , B _A3 and B _{A4 of} the reference area A arranged in such a manner (with no to-be-counted object still appearing in the reference area A).

If a to-count object or a person penetrates into the reference area A, then a corresponding figure C _{1.1... 1.n of} the target is obtained tes in the form of a cluster segment on the respectively at a time t _{1 ... n} recorded image B _{A1 ... n} .

By sequentially joining or juxtaposing the images B _{A1... N} or the cluster segment image C _{1.1... 1n} along a time axis t, a graphic cluster C ₁ appears on the river image B in a mosaic-like manner _{... n} . A cluster C _{1... N} can be constructed from any number of cluster segment maps C _1.1 .

If a counter object or a person passes the reference area A in a relatively rapid manner, then a cluster C _{1... N} assigned to this counter object or person will be made up of relatively few cluster segment maps C _1.1 and be short accordingly. If, on the other hand, a counter object or a person passes the reference area A in a relatively slow manner, then a cluster C _{1... N} assigned to this counter object or person will become a relatively large number of cluster segment maps C _1.1 put together and be correspondingly long. In the event that a toughened object or a person remains on the reference area A, this would result from the continuous image generation by the camera 2 theoretically an infinitely long cluster C _{1 ... n} .

In 3 the transfer of two persons over the reference area A is shown in a purely exemplary way. A first person identified by a first cluster C ₁ has entered the reference area A at a time t ₁ , parts of this person appearing in a first cluster segment map C _1.1 on a picture B _A1 recorded at the time t ₁ . The first person is still on other pictures B _A1 , B _A2 , B _A3 , B _A4 and B _A5 in the form of cluster segment maps C _1.2 , C _1.3 , C _1.4 and C _1.5 added. As in the illustration according to 3 As can be seen, all associated cluster-segment maps C _1.2 , C _1.3 , C _1.4 and C _1.5 caused by the first person are connected to form a uniform cluster C ₁ .

Furthermore, the flow chart B shows according to 3 in the right middle area identified by a second cluster C2 second person who has entered the reference area A at a time t ₁₀ and on a at this time by the camera 2 taken picture B _{A10 appears} as a cluster segment map C _2.1 . The second person is still recorded on further sequential images B _A11 to B _A18 in the form of cluster segment maps C _2.2 to C _2.9 . Again, all cluster segment maps C _2.1 to C _{2.9 associated} with the second person are connected to form a uniform cluster C ₂ .

Since the second person has moved over the reference area A at a speed that is comparatively low compared to the first person, the second cluster C _{2, which} corresponds to a "crawl track", has a substantially greater length than the first cluster C ₁ .

A corresponding one in the evaluation device 3 The algorithm installed now checks whether a cluster C _{1... n produced} on the flow chart B breaks off again in the direction of the time axis t or is continued by no further cluster segment mappings C _{1.1... 1.n.}

In the present illustrative example, as the first cluster C ₁ ruptures after its aufscheinenden on the image B _A5 fifth cluster segment mapper C _1.5 (at time t ₅₎ and is in sequentially on the image B _A5 following image B _A6 (at time t ₆ ) is no longer displayed.

Such tearing off of the first cluster C ₁ in the direction of the time axis t of the flow image B is performed by the evaluation unit 3 is recognized as a completed passing of the reference area A by the first person or as an event to be counted. Accordingly, an electronic count is triggered and on each one with the evaluation 3 linked volatile or non-volatile storage medium stored.

Likewise, the tearing off of the second cluster C ₂ in the direction of the time axis t of the flow image B, namely in the region of the image B _A18 (at the time t ₁₈ ) or after its ninth cluster segment image C _2.9 from the evaluation unit 3 detected as a completed passing of the reference area A by the second person and in turn triggered a count.

Instead of tearing off a graphical cluster C _{1... N} , it would also be possible for the first appearance of a graphical cluster C _{1... N} or to join two or more contiguous cluster segment representations C _1.1 and C _1.2 as from evaluation 3 To define the event to be counted and to use it for the recognition of toughened objects.

Due to the increased reflection properties of the reference area A with respect to its surrounding area 4 This results in a high contrast contrast on the camera 2 taken pictures B _{A1 ... n} . In particular, the counting system can no longer be irritated by shadow throws and different brightnesses of the counter objects. An inventive counting system is thus able to ensure a high counting accuracy.

Basically, it turns out to efficiency Any type of contrasting of the counting objects is advantageous in the present counting process. Depending on the application or installation location of a counting device according to the invention, it may therefore also be advantageous, instead of a reference region A with respect to the surrounding area 4 increased reflection properties and a reference area A 'with respect to the surrounding area 4 increased absorption properties, eg. B. to arrange a black reference area A '.

In such a case, the toughening objects would not appear as clusters C _{1... N} appearing dark relative to the reference region A on the flow chart B (as in FIG 3 instead of the (black) reference region A ', the counter objects would appear as comparatively bright clusters C _{1... n} . A providence of black or dark reference areas A 'may be particularly useful where particularly bright or reflective toughening objects are to be expected in the respective Observanzbereich.

Likewise peculiar to the expression the special contrast properties of the reference area A also suitable high-contrast patterns such. B. checkerboard or strip Patterns and relief-like textures.

In the event that in the respective Observanzbereich the counting device according to the invention already an object with respect to the surrounding area 4 is significantly increased reflection or absorption properties (eg., A chrome fold of a vehicle door) is present, so this object can be used as a reference area A. Another example of the use of an already existing in the Observanzbereich object with respect to the surrounding area 4 significantly increased contrast properties would be z. B. the use of a door or portal threshold or -zarge as reference area A.

In a further development of the counting device according to the invention, both a reference region A with respect to the surrounding region can also be used 4 increased reflection properties, as well as a reference area A 'with respect to the surrounding area 4 increased absorption properties, eg. B. a strip-shaped white reference area A and a strip-shaped black reference area A ', as in 4 represented, one behind the other, wherein a correspondingly coordinated evaluation of the flow image B by the evaluation 3 is made in accordance with the manner described above. Since by means of such an arrangement z. B. persons dressed in dark clothes (which would possibly not be identifiable on an exclusively dark reference area A ') as well as persons who are particularly light or clothed in reflective clothing (which would possibly be unidentifiable on an exclusively bright reference area A'), results in an even higher counting accuracy of the counting method according to the invention.

In the course of a further development of the counting method according to the invention, it is also possible that in one with the evaluation 3 linked database detection parameters are kept, which serve an assignment or classification of appearing on the flow chart B clusters C _{1 ... n} . In this way, for. B. persons are distinguished by dogs or luggage carried.

It is furthermore within the scope of the counting method according to the invention that the optical flow direction of a cluster C _{1... N is} detected and the traversal direction of a toughening object frequenting the reference area A is thereby determined. In this respect, z. B. by the camera-based observation of a single reference area A, which z. B. across an entrance portal of a stadium, the number of people entering the stadium and the number of people leaving the stadium are determined.

Since an already created flow image B or the cluster C _{1... N present on it} does not contain any direction information and therefore does not permit an assessment of the passing direction of the recorded toenail object, it is proposed in the present exemplary embodiment that, parallel to the described assembly, the information from the camera 2 recorded images B _{A1 ... n} also a permanently active flow algorithm runs, which z. B. by means of vector generation, the optical flow direction of an appearing toughened object or the corresponding with this cluster segment maps C _{1.1 ... 1.n} determined.

It should be noted that the merging or the juxtaposition of the images B _{A1... N} resulting in the flow image B can take place in any desired direction. Preferably, a rotating frame buffer system is applied, with a currently captured image B _{A1 ... n being added to} one (eg lower) end of the flow image B, while in return an image B _{A1 ... n} already taken at an earlier time is added _{. n is} removed or deleted again at the other (eg top) end of the flowchart B. In this way, a memory-economical data processing is possible, since in a respective working memory of the evaluation device 3 only a limited number of images B _{A1 ... n} must be kept stored, while those not relevant for a cluster generation images B _{A1 ... n} can be deleted. However, this means that for the already described special case of a verhar of a toughened object on the reference area A and thus of a cluster C _{1... n} that does not break off over a long period of time, this cluster C _{1... n} or the cluster segment mappings C _{1.1. 1.n} continue to follow by means of a "tracking" method and all newly appearing on the images B _{A1 ... n} cluster segment maps C _{1.1 ... 1.n} just that initially detected cluster C _{1 ... n} assign.

It may also be necessary on the camera 2 recorded images B _{A1 ... n} appearing cluster segment images C _{1.1 ... 1.n} virtually manipulate or prepare for merging into a single cluster C _{1 ... n} graphically. Depending on the width y of the reference area A, the outer dimensions of the counting objects and the image acquisition frequency of the camera 2 For example, the fact that some of the images B _{A1... n} or the cluster segment maps C _{1.1... 1n} only _contain a small portion of the to-be-object or person entering the reference area A, eg , B. show a toe. In a mere concatenation of such images B _{A1 ... n} or cluster-segment images C _{1.1 ... 1.n} would be sometimes distances or wave effects between associated pictures segment C _{1.1 ... 1.n} result which could lead to an irritation of the counting system.

For this reason, it may be provided that the cluster segment images C _{1.1... 1n appearing} on the images B _A1 3 due to spatial and / or temporal affinity as belonging together and assigned to a respective toggle object or a cluster C _{1 ... n} , where those mappings C _{1.1 ... 1.n of} this toggle object, which is an interruption to a contiguous portion of the associated Clusters C _{1 ... n} , either merged with this cluster C _{1 ... n} in a virtual manner or be filtered out of the flowchart B. The definition of the spatial / temporal affinity of cluster-segment mappings C _{1.1... 1.n} is determined on the basis of parameters which are determined, inter alia, by the respective application or by the respective location of use of the counting device or by the type of expected counting objects are dependent.

To increase the performance of the counting system, it would also be possible to have multiple cameras 2 use. This may be advantageous above all in the case of observance areas or reference areas A with a very long length extension.

Claims (12)

Method for the contactless counting of toughened objects, preferably of persons by means of a camera-supported evaluation device ( 3 ), characterized in that - at least one reference region (A) is provided, which has higher contrast characteristics than a to-be-detected toughening object than an area adjacent to the reference region (A) ( 4 ), - wherein the reference region (A) of a light source ( 1 ) - and a camera ( 2 ) and an associated evaluation device ( 3 ) is provided, which at consecutive times (t _{1 ... n} ) images (B _{A1 ... n} ) of the reference area (A) creates - and the images (B _{A1 ... n} ) along a time axis (t) to form a flow chart (B) - whereby images (C _{1.1 ... 1.n} ) of a _{toughening object appearing} on the images (B _{A1 ... n} ) along the time axis (t) in each case to one on the flow chart (B) illustrated graphical cluster (C _{1 ... n)} are connected to one another - and wherein a tear-off or a graphical appearance of a cluster (C _{1 ... n)} in the direction of the time axis (t) of the flow of the image (B) as a passing through the reference region (A) is detected by a counter object or as an event to be counted. Method according to Claim 2, characterized in that the evaluation device ( 3 ) detects the flow direction of the graphic clusters (C _{1... n} ) and thereby recognizes the passing direction of the counting objects crossing the reference area (A). Method according to claim 1 or 2, characterized in that the camera ( 2 ) at least a portion of the region adjacent to the reference region (A) ( 4 ) on the images (B _{A1 ... n} ), wherein the position of on these images (B _{A1 ... n} ) appearing boundary edges of the reference region (A) of the evaluation device ( 3 ) is clearly located. Method according to one of Claims 1 to 3, characterized in that the images (C _{1.1 ... 1.n} ) _appearing on the images (B _{A1 ... n} ) are recorded by the evaluation device ( 3 ) are recognized as belonging together on the basis of spatial and / or temporal affinity and assigned to a respective toggle object or cluster (C _{1... n} ), those mappings (C _{1.1... 1.n} ) of this toggle object being an interruption to a contiguous portion of the associated cluster (C _{1 ... n} ), either merged with this cluster (C _{1 ... n} ) in a virtual way or filtered out of the flow picture B. Device for carrying out a contactless method for counting toughened objects by means of a camera-supported evaluation device ( 3 ) according to one of claims 1 to 4, characterized in that the device detects the following components: At least one reference region (A) which has higher contrast properties than a to-be-detected to-be-measured object than an environmental region adjacent to the reference region (A) ( 4 ) - a light source illuminating the reference area (A) ( 1 ) - a camera aimed at the reference area (A) ( 2 ), by means of which at successive times (t _{1 ... n} ) images (B _{A1 ... n} ) of the reference area (A) can be created - one with the camera ( 2 ) associated evaluation device ( 3 ) - which is able to combine the images (B _{A1 ... n} ) along a time axis (t) into a flow image (B) and images (C _{1.1 ... 1.n} ) of one on the images (B _{A1 ... n} ) to be connected to one another along the time axis (t) to form a graphical cluster (C _{1... n} ) shown on the flow chart (B) - the evaluation device ( 3 ) generates a count pulse when tearing or emerging a graphical cluster (C _{1 ... n} ) along the time axis (t) of the flowchart (B). Device according to Claim 5, characterized in that the increased contrast properties of the reference region (A) are provided by providing a reference region (A) with respect to the environment ( 4 ) increased reflection properties can be achieved. Apparatus according to claim 5, characterized in that the increased contrast properties of the reference region (A) by providing a reference region (A ') with respect to the environment ( 4 ) increased absorption properties can be achieved. Device according to one of claims 5 to 7, characterized that at least one reference region (A) with increased reflection properties and at least one reference region (A ') having enhanced absorption properties arranged one behind the other. Device according to one of claims 5 to 8, characterized the reference area (A) is strip-shaped. Device according to one of claims 5 to 9, characterized in that the reference area (A) at the bottom and the light source ( 1 ) is arranged above the floor or above the counting objects to be counted. Device according to one of claims 5 to 9, characterized in that as a reference region (A) already in an observance range of the camera ( 2 ) existing object with respect to the environment ( 4 ) increased reflection or absorption properties is used. Using a frame or a squadron of a door or a Portals as a reference area (A) in a method or a device according to one of preceding claims.