JP2006107060A - Room entering and leaving detector - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a room entering and leaving detector capable of accurately and easily carrying out the detection of the entering and leaving a moving body to and from a room at high speed without being accompanied by the data processing of a large processing amount such as the area calculation of the moving body to be detected when entering and leaving the room. <P>SOLUTION: The room entering and leaving detector is provided with: a difference detection part 101 for obtaining difference data of processing image data inputted from an image input device 106 and reference image data inputted and held beforehand; an image processing part 102 for carrying out a prescribed data analysis on the basis of the difference data and extracting analysis data; a position detection part 103 for detecting the position of the moving body on the basis of the analysis data; a motion detection part 104 for detecting the motion of the moving body on the basis of the analysis data of the processing image data processed almost simultaneously, or position information; and a room entering and leaving determination part 105 for determining movement in an entry direction, movement in an exit direction and U-turn movement between the room entering and leaving directions within the imaging range of the image input device 106 on the basis of the position information and movement information of the moving body and determining the room entering and leaving of the moving body. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to an entry / exit detection device for detecting the entry / exit of an animal body by detecting the movement of an animal body including a human being.

  As a technique for recognizing the movement of an animal body, there is one disclosed in Patent Document 1 below. In the prior art disclosed in this patent document 1, a density projection algorithm is used for recognition of an animal body (particularly, a person), and the face position of the person is detected.

  FIG. 5 shows a system configuration diagram of the person recognition apparatus in the prior art, and FIG. 6 shows a flowchart of the recognition processing operation.

  5 is a CCD camera that captures a person and outputs an image signal 1, an image memory 2 that stores the image signal 1 for a given frame, an image signal 1 and an image memory that are input after n frames. 2 based on the image signal recorded in advance 2, that is, an absolute difference unit 3 that obtains an absolute difference between the two images, that is, a motion region in both images and outputs the difference value, and an image signal 1 based on the image signal 1. A threshold value determination unit 4 for determining a threshold value for binarizing, and a comparison unit 5 that inputs a difference value from the absolute difference unit 3 and performs binarization processing using the threshold value determined by the determination unit 4 Is done.

  Further, the person recognition apparatus includes an X-axis density projection unit 6, a maximum X-axis projection coordinate extraction unit 7, a Y-axis density projection unit 8, a maximum Y-axis projection coordinate extraction unit 9, a face feature extraction / determination unit 11, and edge detection. Unit 12, output generation unit 14, head width extraction unit 19, moving object area calculation unit 31, moving object area calculation / memory unit 41 of the previous frame, and comparison unit 42. Each of these parts will be described with reference to FIG.

  The head width extraction unit 19 is activated only in response to the comparison result signal 20 given from the comparison unit 42 described later, and the X coordinate (Xa2, Xb2) of the head width of the human body in the detected moving body region. ) The moving object area calculation unit 31 is a part that calculates the projected area of the moving object.

  The moving object area calculation / memory unit 41 of the previous frame calculates the moving object area for each frame based on the obtained coordinates Xa1, Xb1, and Ytop, and sequentially stores the calculated area values. The comparison unit 42 compares the moving object area of the current frame output from the calculation unit 31 with the moving object area of the previous frame stored in the memory unit 41, and the current frame is greater than the previous frame. In response to this, the comparison result signal 20 is output, and the head width extraction unit 19 is activated. Note that the moving body area calculation / memory unit 41 of the previous frame clears the area of the previous frame of the memory unit 41 each time the stored moving body area of the previous frame is read (comparison completion by the comparison unit 42), Repeat the process of storing the current calculated area.

  Next, the operation of the person recognition apparatus in FIG. 5 will be described with reference to FIG. Using the image signal 1 obtained by imaging with a CCD camera or the like, the coordinates Xa1, Xb1 and Ytop are obtained in the same manner as described above by the image memory 2 to the maximum Y-axis projected coordinate extraction unit 9 (S1 in FIG. 6). To S6). Next, the moving object area calculation unit 31, the moving object area calculation / memory unit 41 of the previous frame, and the comparison unit 42 detect whether the moving object image area of the current frame is larger than that of the previous frame. In response, the comparison result signal 20 is output and the head width extraction unit 19 is activated. The head width extraction unit 19 is activated by the comparison result signal 20 only when the moving object image area of the current frame is larger than that of the previous frame, and extracts the head width in the same manner as described above, Output to the feature extraction / determination unit 11. The face feature extraction / determination unit 11 uses the coordinates Xa2 and Xb2 output from the head width extraction unit 19 to determine whether facial features are sufficiently extracted in the head width of the animal body, and outputs a face determination result signal 13 The output generation of the output generation unit 14 is controlled.

As described above, according to the person recognition apparatus of FIG. 5, the moving object approaching the camera is limited by the moving object area calculating unit 31, the moving object area calculating / memory unit 41 of the previous frame, and the comparing unit 42. It can be done, and approaching means that the face is almost facing the front, so that the detection accuracy in the person recognition device can be improved.
JP-A-7-73298

  However, in the above-described conventional person recognition device, in order to detect the face position, it is necessary to obtain the X-axis density projection, the Y-axis density projection, and even the area. There is a drawback of too much throughput. In addition, in entering and leaving the room, there is a case where the moving object that has entered the room once makes a U-turn and leaves the room, so a new algorithm is considered necessary.

  The present invention has been made in view of the above-described problems, and its object is to easily and rapidly enter an animal without involving a large amount of data processing such as calculating the area of the object to be detected in entry / exit. An object of the present invention is to provide an entrance / exit detection device capable of accurately executing exit.

  In order to achieve the above object, an entrance / exit detection device according to the present invention is an entrance / exit detection device that detects the entrance / exit of an animal body by detecting the movement of an animal including a person, and is input from an image input device. A difference detection unit that obtains difference data between the processed image data and the reference image data that was input and held before the input of the processed image data, and a predetermined data analysis process is executed based on the difference data An image processing unit for extracting analysis data; a position detection unit for detecting the position of the moving object based on the analysis data; and the analysis data or the position of the processed image data processed in tandem with each other A motion detection unit that detects a movement of the moving object based on the information; a position information detected by the position detection unit of the moving object; and a motion information detected by the motion detection unit. An entry / exit determination unit that distinguishes between movement in the entry direction, movement in the exit direction within the imaging range, and U-turn movement between the entry and exit directions, and determining entry / exit of the moving object. Features.

  In the entrance / exit detection device having the above-described characteristic configuration, it is also preferable that the image processing unit performs noise removal by expansion processing or contraction processing on the differential data before executing the data analysis processing.

  In the entrance / exit detection device having the above-described characteristic configuration, the image processing unit preferably extracts the analysis data by performing density projection on the difference data as the data analysis processing. Furthermore, it is also preferable that the image processing unit binarizes the difference data before performing density projection on the difference data. The image processing unit may perform density projection on a partial area of the difference data when performing density projection on the difference data.

  In the entrance / exit detection device having the above-described characteristic configuration, the image processing unit preferably extracts the analysis data by performing a labeling process on the difference data as the data analysis process.

  In the entrance / exit detection device having the above-described characteristic configuration, the motion detection unit holds only information on the position and length of the moving object in the analysis data of the processed image data processed immediately before. preferable.

  Further, the entrance / exit detection device having the above-described characteristic configuration is configured so that the position detection unit does not recognize the object as the moving object when the length of a certain object obtained from the analysis data is equal to or less than a predetermined threshold. Also good.

  In the entrance / exit detection device having the above-described characteristic configuration, when the length of a certain object obtained from the analysis data is equal to or greater than a predetermined threshold, the position detection unit recognizes the object as a collection of moving objects. You may do it.

  The entrance / exit detection device having the above-described characteristic configuration may be configured such that the processed image data and the reference image data input from the image input device are captured in a direction parallel to the bottom surface of the entry / exit detection target area. The entry and exit directions may be parallel to the bottom surface.

  Further, in the entrance / exit detection device having the above-described characteristic configuration, the processed image data and the reference image data input from the image input device are image data taken in a vertical direction toward the bottom surface of the entry / exit detection target area. There may be a configuration in which the entry and exit directions are any one of a vertical direction, a horizontal direction, and an oblique direction on the screen.

  In the entrance / exit detection device having the above-described characteristic configuration, it is also preferable that the difference detection unit uses image data in which the moving object does not exist in the screen as the reference image data.

  In the entrance / exit detection device having the above-described characteristic configuration, it is preferable that the difference detection unit uses the processed image data input immediately before as the reference image data.

  In addition, when the image input from the image input device is an RGB color image, the entrance / exit detection device having the above characteristic configuration has a particularly large luminance change in the color image as the processed image data and the reference image data. You may make it use G component.

  In the entry / exit detection device having the above-described characteristic configuration, it is preferable that the entry / exit determination unit does not determine that the moving object moving in the U-turn is entering or leaving the room.

  According to the entrance / exit detection device according to the present invention having the above-described characteristic configuration, there is no need for two-dimensional data processing such as area calculation, and only data analysis processing with a small one-dimensional processing amount such as density projection and labeling processing is performed. Because the entry / exit detection is performed at the, entry / exit of the moving object can be detected at high speed. Even when the moving object moves U-turns, it can be excluded or distinguished from entering or leaving the room by detecting the direction of travel.

  Next, an embodiment of the entrance / exit detection device according to the present invention (hereinafter, appropriately referred to as “the present invention device”) will be described with reference to the drawings.

  FIG. 1 is a system configuration diagram showing a configuration example of the device 100 of the present invention. As shown in FIG. 1, the apparatus of the present invention includes a difference detection unit 101, an image processing unit 102, a position detection unit 103, a motion detection unit 104, and an entrance / exit determination unit 105.

  In this embodiment, the image input device 106 such as a video camera sets the detection target area for detecting the movement of a moving object such as a person in a direction (horizontal) parallel to the bottom surface (floor surface or ground) of the area. The moving object is assumed to move in the horizontal direction (left and right direction on the imaged screen) on the bottom surface of the area, and as a condition, the moving object is left and right within the imaging range. It is assumed that the room does not enter from both directions at the same time (not in a matching state), and that within the imaging range, a plurality of moving objects do not overlap with the camera when entering or leaving the room. However, within the imaging range, multiple moving objects enter the same direction, one moving object overtakes another moving object, makes a U-turn after passing, and enters the imaging area from the left and right sides. After that, it is okay to make a U-turn and leave the room.

  The difference detection unit 101 holds the image data input from the image input device 106, the most recently input processed image data of the input image data, and the reference that has been input and held before that. Difference data with image data is generated. In the “initial image input mode”, the reference image data is captured by the image input device 106, input to the difference detection unit 101, and the initial image stored in the image memory in the difference detection unit 101 is used. In the “initial image input mode”, the operation after obtaining the reference image data is not performed. After changing the mode from the “initial image input mode” to the “operation mode”, the difference detection unit 101 makes an absolute difference between the reference image data acquired in the “initial image input mode” and the latest input processed image data. Difference data obtained by calculation is generated and output to the image processing unit 102.

  The image processing unit 102 executes predetermined data analysis processing based on the difference data, extracts the analysis data, and outputs it to the position detection unit 103. Specifically, for example, analysis data is extracted by performing density projection. More specifically, for example, when the image input device 106 images the detection target area in the horizontal direction with respect to the floor surface, and the moving object moves horizontally in the detection target area, the horizontal direction is the X-axis direction. In this case, the density projection on the X axis is extracted as analysis data.

  The position detection unit 103 detects the position of the moving object based on the extracted analysis data. In the case of density projection, for example, when the density projection value continuously exists over a certain width on the X axis, it is determined that there is an animal in the corresponding X coordinate range. This is shown in FIG. As shown in FIG. 2, when the width at which the density projection value is equal to or greater than a predetermined threshold is equal to or greater than a certain value, the object is recognized as an animal body. Position information (X coordinate) when recognized as a moving object is output to the motion detection unit 104.

  The motion detection unit 104 stores the position information of the moving object based on the analysis data of the processed image data processed before and after the time, and detects the movement of the moving object from the previous position information. Specifically, the left and right X coordinates of the X coordinate range recognized as a moving object are sequentially stored. The frame number at the time of the immediately previous processing is (n-1), the frame number currently being processed is n, and the width of the moving object between the two processed image data of frame (n-1) and frame (n) ( Whether the left end and the right end (X coordinate difference) has increased or decreased is detected and output to the entrance / exit determination unit 105.

  The entrance / exit determination unit 105 captures an image of the image input device 106 based on the position information (X coordinate) detected by the position detection unit 103 of the moving object and the motion information (width increase / decrease) detected by the motion detection unit 104. The movement of the entry direction within the range, the movement in the exit direction, and the U-turn movement between the entry and exit directions are distinguished, and the entry and exit of the moving object is determined, and the moving object passes through the imaging range from right to left. The case and the case of passing from left to right are counted as one entering the room and leaving the other.

  Next, with reference to FIG. 3, the entrance / exit detection operation of the apparatus of the present invention in the position detection unit 103, the motion detection unit 104, and the entrance / exit determination unit 105 will be described in more detail. FIG. 3 shows a flowchart of the entry / exit detection operation.

  In the moving object position detection step (# 10), it is detected whether the moving object is at the left end, the right end, or the center of the imaging range. For example, as shown in FIG. 4, the X coordinate on the left side of the moving object is Xs, the X coordinate on the right side is Xe, the threshold value on the left end is XLth, the threshold value on the right end is XRth, and the inequality shown in Equation 1 is When it is established, it is defined that the moving object is at the left end, and when the inequality shown in Equation 2 is satisfied, it is defined that the moving object is at the right end, and other than that, the moving object is defined at the center.

(Equation 1)
Xs <XLth

(Equation 2)
Xe> XRth

  In addition, although the case where both the said inequality of the said number 1 and number 2 is materialized can also be assumed, it is a case where a moving body exists in the whole imaging range, and this case shall be excluded in this embodiment.

  Next, in step # 20, the subsequent processing is branched depending on the three existing positions of the moving object determined by the above definition. If it exists in the central portion that is neither the left end nor the right end, the subsequent motion detection and entry / exit determination are not processed. If the moving object is at the left end, the process proceeds to step # 30 and subsequent steps. If the moving object is at the right end, the process proceeds to step # 40 and subsequent steps.

  First, the case where the moving object is at the left end in the determination of step # 20 will be described. In step # 30, a change in the stored detection width one frame before is determined. If this value is unchanged, the process proceeds to step # 31. Any other increase or decrease is not processed because it has been processed one frame before.

  In step # 31, the change in the width of the moving object is calculated. If the width increases, the process proceeds to step # 32. If the width decreases, the process proceeds to step # 33. If there is no change in width, it is determined that there is no movement or that the moving object has not reached the left end, and the process is terminated and the process ends.

  When the process proceeds to step # 32, since the position of the moving object is at the left end and the width of the moving object has increased, it is determined that the moving object has entered the room, and the “enter from left” mode is set.

  When the process proceeds to step # 33, it is determined that the moving object has left to the left because the position is at the left end and the width of the moving object has decreased. In this case, the process further proceeds to step # 34.

  In step # 34, if the “entry from right” mode is set for the moving object to be determined, the process proceeds to step # 35, and if not (the “entry from left” mode is set). If YES, the process proceeds to step # 36.

  In Step # 35, since entering the room from the right and leaving the room to the left, "passing counting from the right to the left" is performed. In Step # 36, it is determined that the moving object enters the room from the left and leaves the room to the left, and does not count.

  Next, the case where the moving object is at the right end in the determination of step # 20 will be described. In step # 40, a change in the stored detection width one frame before is determined. If this value is unchanged, the process proceeds to step # 41. Any other increase or decrease is not processed because it has been processed one frame before.

  In step # 41, the change in the width of the moving object is calculated. If the width increases, the process proceeds to step # 42, and if it decreases, the process proceeds to step # 43. On the other hand, if there is no change in width, it is determined that there is no movement or the moving object has not reached the right end, and the process is terminated and the process is terminated.

  When the process proceeds to step # 42, since the position of the moving object is at the right end and the width of the moving object has increased, it is determined that the moving object has entered the room, and the "enter from right" mode is set.

  When the process proceeds to step # 43, it is determined that the moving object has left the room because the position is at the right end and the width of the moving object has decreased. In this case, the processing is further moved to step # 44.

  In step # 44, if the “entry from left” mode is set for the object to be determined, the process proceeds to step # 45, and if not (the “entry from right” mode is set. If YES, the process proceeds to step # 46.

  In step # 45, since entering the room from the left and leaving the room to the right, “passing counting from the left to the right” is performed. In Step # 46, it is determined that the moving object enters the room from the right and leaves the room to the right, and does not count.

  With the above processing, the entry / exit can be determined by distinguishing “passing from right to left”, “passing from left to right”, and “U-turn”. For example, if “passing from right to left” is a movement in the entry direction and “passing from left to right” is a movement in the exit direction, entry / exit can be distinguished and counted by each count.

  The operation described with reference to the flowchart of FIG. 3 is performed for each frame of input processed image data.

  Next, another embodiment of the device of the present invention will be described.

  <1> In the above embodiment, before the image processing unit 102 executes data analysis processing such as density projection based on the difference data, the difference data is subjected to contraction processing several times to remove minute noise. Furthermore, it is also preferable to perform a pretreatment to fill the hole or the like when the animal body is opened by the expansion process so as to ensure the shape of the animal body.

  <2> In the above embodiment, the position detection and width detection of the moving object by density projection have been described as an example of the data analysis process by the image processing unit 102. However, instead of density projection, labeling processing is performed on difference data. The position of the moving object and the width are detected based on the width of the pixel set having a predetermined label value in the X-axis direction, and the entrance / exit of the moving object may be determined based on the change in the width. Absent.

  <3> In the above embodiment, before the image processing unit 102 executes data analysis processing such as density projection based on the difference data, the difference data may be binarized once. By such preprocessing, even when the difference data is very small, the position of the moving object can be accurately detected in data analysis processing such as density projection.

  <4> In the above-described embodiment, the difference data for which the image processing unit 102 performs data analysis processing such as density projection may target not only the entire imaging range but also a partial area thereof. For example, data in the vicinity of the bottom surface (floor surface or ground) or top surface (ceiling, etc.) of the imaging range is excluded from the data analysis processing targets. As a result, it is possible to remove data components unnecessary for detecting the position of the moving object.

  <5> In step # 31 and step # 41 of the above embodiment, analysis data such as density projection one frame before is stored in order to determine a change in detection width in the processed image data of the current frame. When determining the change in detection width based on the previous analysis data and the analysis data in the processed image data of the current frame, the analysis data of the previous frame is not information on the entire image data such as density projection, Only the information on the position and width (length) of the moving object may be stored and used for processing the next frame.

  <6> In the above embodiment, the description has been made assuming that the moving object recognized by the position detection unit 103 is a single moving object. However, the density projection value exists continuously over a certain width, and the length is long. Is greater than or equal to a predetermined threshold value, it is determined as an assembly of a plurality of moving objects, and when the entrance / exit determination unit 105 counts entrance / exit, the concentration projection value continuously exceeds a certain width. In another preferred embodiment, the value obtained by dividing the existing length by the predetermined threshold value is regarded as the number of the aggregates, and the number is counted.

  <7> In the above embodiment, when the moving object passes from the right to the left and from the left to the right within the imaging range, the entrance and the exit are counted separately. Although the case where the count is not performed has been described, in the case of the U-turn, the count may be independently made distinct from the others.

  <8> In the above embodiment, the image input device 106 images the detection target area in a direction (horizontal direction) parallel to the bottom surface (floor surface or ground surface) of the area, and the moving object Although it is assumed that the image moves on the bottom surface in the horizontal direction (left and right on the imaged screen), the image input device 106 is installed at a high position (such as the ceiling) of the detection target area, and the area is moved toward the bottom surface. The image data photographed in the downward direction may be used as the processed image data and the reference image data. In this case, the moving direction (entrance and exit directions) of the moving object may be any of an up-down direction, a left-right direction, and an oblique direction on the screen.

  When the moving object moves in the vertical direction on the screen, the density projection for the difference data is performed on the Y axis parallel to the moving direction. When the moving object moves in the left-right direction on the screen, the density projection for the difference data is performed on the X axis parallel to the moving direction. Further, when the moving object moves diagonally on the screen, the density projection for the difference data is performed on the axis having the larger component of the moving vector of the moving object moving in the diagonal direction.

  <9> In the above embodiment, when the difference detection unit 101 receives input of image data from the image input device 106, two modes of “initial image input mode” and “operation mode” are provided. ”, The reference image data is input, but the image data input mode is not distinguished, that is, the“ initial image input mode ”and the“ operation mode ”are not provided, and always one frame before. The input processed image data may be handled as reference image data. In this case, for example, when the moving object moves in the left-right direction on the screen, difference data in which the left and right edges of the moving object are emphasized is generated, so the image processing unit 102 matches the characteristics of the difference data. Data analysis processing may be executed.

  <10> In the above embodiment, the image data captured by the image input device 106 and input to the difference detection unit 101 is subject to data processing, so that the image data is sufficient as a monochrome image. When the image captured by the image input device 106 is an RGB color image, data analysis processing such as density projection is performed using a G component having a particularly large luminance change in the RGB signal. Is preferred.

The block block diagram which shows the system structural example in one Embodiment of the entrance / exit detection apparatus which concerns on this invention The figure which showed the mode of the position detection of the moving body by density | concentration projection in one Embodiment of the entrance / exit detection apparatus which concerns on this invention It is a flowchart which shows the entrance / exit detection operation in one Embodiment of the entrance / exit detection apparatus which concerns on this invention. The figure explaining the determination method of the position of the moving object in the imaging range in one Embodiment of the entrance / exit detection apparatus which concerns on this invention The block diagram which shows the system configuration | structure of the conventional person recognition apparatus The flowchart which shows the recognition processing operation | movement of the conventional person recognition apparatus shown in FIG.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100: Entrance / exit detection apparatus concerning this invention 101: Difference detection part 102: Image processing part 103: Position detection part 104; Motion detection part 105: Entrance / exit determination part 106: Image input device

Claims (15)

An entry / exit detection device for detecting the entry / exit of the moving object by detecting the movement of the moving object including a human,
A difference detection unit for obtaining difference data between the processed image data input from the image input device and the reference image data input and held before the input of the processed image data;
An image processing unit that executes predetermined data analysis processing based on the difference data and extracts analysis data;
A position detector that detects the position of the moving object based on the analysis data;
A motion detection unit that detects the motion of the moving object based on the analysis data or the position information of the processed image data processed before and after in time;
Based on the position information detected by the position detection unit of the moving object and the movement information detected by the motion detection unit, movement in the entrance direction, movement in the exit direction within the imaging range of the image input device, and An entrance / exit determination unit that distinguishes U-turn movement between the entrance and exit directions and determines entrance / exit of the moving object,
An entrance / exit detection device comprising:   The entrance / exit detection device according to claim 1, wherein the image processing unit performs noise removal by an expansion process or a contraction process on the difference data before executing the data analysis process.   The entrance / exit detection device according to claim 1, wherein the image processing unit extracts the analysis data by performing density projection on the difference data as the data analysis processing.   The entrance / exit detection device according to claim 3, wherein the image processing unit binarizes the difference data before performing density projection on the difference data.   The entrance / exit detection device according to claim 3 or 4, wherein the image processing unit performs density projection on a partial region of the difference data when performing density projection on the difference data. .   The entrance / exit detection device according to claim 1, wherein the image processing unit extracts the analysis data by performing a labeling process on the difference data as the data analysis process.   The said motion detection part hold | maintains only the information regarding the position and length of the said moving body in the said analysis data of the said process image data processed immediately before, The any one of Claims 1-6 characterized by the above-mentioned. The entry / exit detection device according to the item.   The said position detection part does not recognize the said object as the said moving body, when the length of a certain object obtained from the said analysis data is below a predetermined threshold value, The any one of Claims 1-7 characterized by the above-mentioned. The entrance / exit detection device described in 1.   The position detection unit recognizes the object as a collection of moving objects when the length of a certain object obtained from the analysis data is equal to or greater than a predetermined threshold. The entrance / exit detection apparatus of Claim 1.   The processed image data and the reference image data input from the image input device are image data photographed in a direction parallel to the bottom surface of the entrance / exit detection target area, and the entrance and exit directions are the The entrance / exit detection device according to claim 1, wherein the entrance / exit detection device is in a direction parallel to the bottom surface.   The processed image data and the reference image data input from the image input device are image data photographed in a vertical direction toward the bottom surface of the entry / exit detection target area, and the entrance and exit directions are on the screen. The entrance / exit detection device according to any one of claims 1 to 9, wherein the entry / exit detection device is any one of a vertical direction, a horizontal direction, and an oblique direction.   The entrance / exit detection device according to any one of claims 1 to 11, wherein the difference detection unit uses image data in which the moving object does not exist in a screen as the reference image data.   The entrance / exit detection device according to claim 1, wherein the difference detection unit uses the processed image data input immediately before as the reference image data.   The G component of the color image is used as the processed image data and the reference image data when the image input from the image input device is an RGB color image. The entrance / exit detection apparatus of Claim 1.   The entrance / exit detection unit according to any one of claims 1 to 14, wherein the entrance / exit determination unit does not determine that the moving object moving in the U-turn is entering or exiting.

Images