JP2011248836A - Residence detection system and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a technology useful for business management and the like, which detects a location with a high degree of stillness, residence, and/or cluster of a person in an area of an imaging range, using a moving image taken by a fixed camera.SOLUTION: A residence detection system 10 performs: with either moving image data of an imaging range 45 taken by a surveillance camera 30 or analysis information 24 (person and motion line information 102) on such data, and matrix data 62 including a plurality of elements made by dividing a frame area into a plurality of blocks, mapping information on a person or a motion line in a matrix and recording a score of each element corresponding to a block where the person or the motion line overlaps; extracting the element with a high score; and outputting information on the block of the element with a high score, as detection result information on a residence location.

Description

  The present invention relates to a technique for performing data information processing using a moving image captured by a fixed camera such as a surveillance camera, and in particular, a person or the like stays in a region of a moving image captured in a facility such as a store. The present invention relates to a technique that is effective when applied to a system and program for detecting a location.

  Currently, surveillance cameras and surveillance camera systems (hereinafter sometimes referred to simply as “surveillance cameras”) have been installed for various purposes such as crime prevention in various facilities such as convenience store stores. Yes. Surveillance cameras continue to be highly functional and high performance, and cameras having various functions are provided to the market.

  As a function of the surveillance camera, in addition to the basic function of shooting and recording images inside and outside the facility (moving images), as an additional function, a moving object is selected based on image analysis processing on the captured moving images. There are a function to detect and a function to detect a specific object such as a person or a face based on the detected feature of the moving object.

  Japanese Patent Laying-Open No. 2010-15465 (Patent Document 1) describes a system that measures the number of visitors to a store or the like using image data of a surveillance camera.

  Japanese Patent Laid-Open No. 2007-74330 (Patent Document 2) analyzes whether or not a customer has a willingness to purchase, whether or not a suspicious person, etc. It describes the technology to recognize. In Patent Document 2, the store PC specifies the monitoring target person from the characteristics of the person included in the image captured by the monitoring camera, and measures the staying time from the appearance of the monitoring target person to the present time. It is described that the person in charge is notified according to the determination. In addition, it is described that the age group and gender of the monitoring subject are estimated.

  Japanese Patent Laying-Open No. 2006-197226 (Patent Document 3) describes a surveillance camera system that can be used not only for crime prevention purposes but also for store management. The use of a fixed camera and a movable camera is described.

JP 2010-15465 A JP 2007-74330 A JP 2006-197226 A

  Conventional surveillance camera systems are often installed and used for crime prevention purposes. For security purposes, it is possible to roughly grasp a person in a store and is effective. As a purpose other than the crime prevention purpose, it may be used for the purpose of business management or the like in a facility such as a convenience store store (for example, Patent Documents 1 to 3), but the utilization is insufficient.

  For example, in a conventional surveillance camera system, the possibility of quantifying and detecting the situation such as stillness, staying, and gathering of people in the area of the shooting range and using them for business management has not been sufficiently studied and realized. For example, in the background area corresponding to the shooting range (for example, the arrangement configuration of products in the store), the customer visits or stops at a specific location (passage, product shelf, etc.), or multiple A system for determining a specific location (“staying location”) due to the gathering and passing of people, etc. has not been realized. For example, Patent Literature 1 is measurement of the number of persons, and Patent Literature 2 is measurement of stay time, and does not determine a specific staying place in the store.

  In view of the above, an object of the present invention is to detect a staying location by using a moving image taken by a fixed camera such as a surveillance camera to quantify the degree of stillness, staying, and gathering of a person in the shooting range area. It is to provide a system that can be used for business management. The above and other objects and novel features of the present invention will be apparent from the description of this specification and the accompanying drawings.

  To achieve the above object, a typical embodiment of the present invention has the following characteristics. This stay detection system is a system built in or connected to a fixed camera system including a fixed camera for shooting a predetermined shooting range, and detects the stay of an object such as a person in the shooting range area. Acquire at least one of moving image or still image frame data to be detected corresponding to the shooting range or image analysis information thereof, and based on the frame data or image analysis information, a person in the frame area corresponding to the shooting range or Using a data acquisition unit that performs processing for acquiring the flow line information and matrix data having a plurality of elements obtained by dividing the frame region corresponding to the shooting range into a plurality of blocks, the person in the frame region or its Map the flow line information to the blocks on the matrix, and the person or its flow line overlaps Residue detection that performs the process of recording the score for each element corresponding to the lock, the process of extracting the element having a high score in the matrix, and the process of outputting the block information of the element having a high score as the detection result information of the staying part Part.

  The present invention can also be applied to a program that causes a computer to function as the system as described above.

  According to a typical embodiment of the present invention, a moving image captured by a fixed camera such as a surveillance camera is used to quantify the degree of stillness / staying / aggregation of a person or the like in a shooting range area. The location can be detected and used for business management.

It is the figure which showed the outline | summary about the structural example of the surveillance camera system containing the residence detection system which is one embodiment of this invention. It is the flowchart which showed the outline | summary about the example of the flow of the stay detection process in one embodiment of this invention. It is the figure which showed the example of the area | region of the imaging | photography range, and correspondence information in one embodiment of this invention. It is the figure which showed the structural example of the matrix corresponding to the area | region of the imaging | photography range, and the example of a plot of the person flow line on a matrix in one embodiment of this invention. It is the figure which showed the example of the element extraction on a matrix, and the example of a residence location in one embodiment of this invention. It is the figure which showed the example of the stay location in one embodiment of this invention.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. In the surveillance camera system including the stay detection system according to the present embodiment, a person (including a case of a plurality of persons) stays and stays in a shooting range area based on a moving image taken by a fixed camera in a facility such as a store. -Determine and detect places ("staying places") where the degree of gathering ("staying degree") is high.

[System configuration]
FIG. 1 shows an outline of a configuration example of a surveillance camera system (fixed camera system) 100 including a system 10 according to an embodiment of the present invention. The surveillance camera system 100 includes a stay detection system 10, a surveillance camera server 20, a surveillance camera (fixed camera) 30, a store 40, and the like.

  One or more surveillance cameras 30 are installed in the store 40. The surveillance camera 30 captures a predetermined imaging range 45. Each surveillance camera 30 is connected to the surveillance camera server 20. The monitoring camera server 20 may be configured inside the store 40 or may be configured outside the store 40 via a network or the like. The stay detection system 10 is connected to the monitoring camera server 20. The stay detection system 10 may be configured as a system in the store 40 or may be configured as a system on an external network.

  The store 40 is, for example, a convenience store store. The imaging range 45 of the monitoring camera 30 is for a predetermined fixed space area such as a passage in the store 40 or a product display shelf. In the present embodiment, the shooting range 45 includes a merchandise display shelf in the store 40 (FIG. 3). A moving image shot for the shooting range 45 by the monitoring camera 30 is recorded in the monitoring camera server 20 as moving image data 23.

  The monitoring camera server 20 includes, for example, a computer system and each unit of the monitoring camera control unit 21 and the image analysis unit 22 implemented by a software program, and each data or database of the moving image data 23 and the analysis information 24. As a result, the monitoring camera server 20 has a function of recording a moving image taken by the monitoring camera 30 as moving image data 23. In addition, the surveillance camera server 20 performs image analysis processing by the image analysis unit 22 based on the recorded moving image data 23, so that a moving person and its face in the area of the shooting range 45, and It has a known function (person detection function and flow line detection function) that can detect the flow line (trajectory) and the like. The processing content of the image analysis unit 22 and the like can be variably set based on the setting information. Analysis information 24 (person / flow line information 102) by the image analysis unit 22 (person detection function and flow line detection function) is recorded in the monitoring camera server 20.

  In the present embodiment, the surveillance camera server 20 has the person detection function and the flow line detection function, and the stay detection system 10 uses the above-described known detection function (its detection information) as a means. (Residence detection) is realized. The stay detection system 10 may be provided with the detection function described above. The detection target is not limited to a person, but can be another moving body (for example, an automobile). In that case, a detection function corresponding to the detection target is provided, and the corresponding coordinate information is processed.

  The monitoring camera server 20 controls the monitoring camera 30 by the monitoring camera control unit 21. This control may include, for example, execution / stop of shooting by the monitoring camera 30, setting or changing the shooting range 45, changing shooting conditions such as exposure, and the like. In particular, it is possible to set a desired shooting range 45 by changing the orientation of the monitoring camera 30 or the like.

  The stay detection system 10 includes, for example, a data acquisition unit (frame processing unit) 11, a stay detection unit 12, and an output unit 13 which are configured by a computer system and implemented by a software program, frame data 60, person / motion. Each data or database includes line information 61, matrix data 62, and detection results (staying information) 63. As a result, the stagnation detection system 10 has a function of calculating and detecting information on the stagnation location (the stagnation information 63) in the area of the photographic range 45 from the photographic moving images obtained by the surveillance camera 30 and the surveillance camera server 20.

  Although not shown, the stay detection system 10 includes setting means for setting various types of information by operations of a store manager or the like, and can set a processing target frame, time, determination threshold value, and the like. .

  The data acquisition unit (frame processing unit) 11 acquires, from the monitoring camera server 20, moving image data 23 and analysis information 24 (person / flow line information 102), which are data information necessary for processing in the stay detection unit 12. ·input. The acquired moving image data 23 is stored as frame data 60 including a plurality of continuous still images (frames), for example. The acquired person / flow line information 102 is stored as person / flow line information 61. The data acquisition unit 11 may perform a process of dividing the frame area into a plurality of blocks (a process related to the creation of the matrix data 62).

  If the data acquisition unit 11 can acquire time-series coordinate information of a person / flow line as the person / flow line information 102 from the monitoring camera server 20, the data acquisition unit 11 acquires or processes the moving image data 23 (frame data 60). Is not required. For example, when the surveillance camera server 20 and the stay detection system 10 (server, etc.) are connected via a network, the data acquisition unit 11 acquires only the person / flow line information 102 (for example, text data format). (Receive) so that the moving image data 23 is not exchanged.

  The stay detection unit 12 uses the frame data 60, the person / flow line information 61, the matrix data 62, and the like to perform stay detection processing (FIG. 2, described later), and stores the detection result (including stay information) 63. To do. The staying information 63 includes information on the staying place detected in the area of the shooting range 45 of the monitoring camera 30. The processing content of the stay detection unit 12 and the like can be variably set based on the setting information.

  The output unit 13 performs output processing using the detection result (retention information) 63. For example, data transmission of the detection result 63 to a computer terminal such as an administrator of the corresponding store 40 and information display on the terminal screen are performed. Further, for example, when displaying the detection result 63 and the like superimposed on the captured video in real time, the output unit 13 feeds back the data of the detection result 63 to the monitoring camera server 20 or the like, and the captured video ( The staying information 63 and the like may be overlaid on the imaging range 45).

  Examples of the frame data 60 and the person / flow line information 61 are the frame 301 (person flow line 310) in FIG. An example of the matrix data 62 is 401 in FIG. Examples of the detection result (retention information) 63 are 502 in FIG. 5 and FIG.

  In this embodiment, an example in which the stagnation detection system 10 is implemented as a device / system independent of the monitoring camera server 20 is shown. However, the program and data constituting each part of the stagnation detection system 10 are monitored. It may be incorporated in the camera server 20 so as to be integrated with the monitoring camera server 20.

[Processing content]
FIG. 2 shows an outline of an example of a flow of processing (staying detection processing) when calculating and detecting a staying location and the like based on the video captured by the monitoring camera 30 in the present system (10, 100). For example, based on the operation on the computer terminal screen by a store manager or the like, the setting process (S01), the photographing process (S02), the stay detection process (S10 to S40), etc. are instructed and executed to the stay detection system 10. Further, in the present embodiment, after photographing the target photographing range 45 in S02, the processing after S10 is performed using the photographing data or the like. However, the processing after S10 is performed in real time with the photographing execution in S02. It can also be set as the form to perform. If the setting process (S01) and the photographing process (S02) have been completed in advance, the process execution can be omitted. When the analysis process (S10) has been completed on the monitoring camera server 20 side, the analysis process (S10) on the stay detection system 10 side can be omitted.

  (S01) In S01, as the setting process, based on the operation of the store manager or the like, the staying detection system 10 (setting means) is set to the shooting range 45, frame, date / time, condition or threshold, matrix ( 62) information such as the division configuration is set.

  (S <b> 02) In S <b> 02, shooting by the monitoring camera 30 is executed for the target shooting range 45, and the data acquisition unit 11 acquires the target moving image data 23 from the monitoring camera server 20 and sets it as frame data 60. Alternatively, the data acquisition unit 11 may acquire the captured moving image data 23 as the frame data 60. For example, the store manager or the like designates the start / end of shooting and acquires an arbitrary number of frames as detection processing targets. In particular, it is also possible to acquire shooting data for each date and set it as a detection processing target.

  (S10) In S10, processing (S11, S12) for detecting a person and its flow line from the moving image (frame region) in the shooting range 45 of the monitoring camera 30 is performed. In the present embodiment, the detection is performed using a known detection function of the image analysis unit 22 provided in the monitoring camera server 20 described above.

  (S11) First, a person is detected from a frame region (for example, 301 in FIG. 3) using the person detection function. If there are multiple persons, each is detected. The detection information includes, for example, information on the coordinates of the current position of the person in the frame area. The coordinates may be, for example, pixel coordinates or position information in units coarser than the pixels, depending on the detection method.

  (S12) For the person detected in S11, the trajectory (flow line) moved in the frame area is grasped in the form of coordinates in the frame area. For example, the flow line (for example, 310) is formed by grasping the coordinates of the current position of the person for each predetermined frame (for example, one frame) and connecting the coordinates.

  In FIG. 3, a frame region 301 is an example corresponding to a three-dimensional space region of a predetermined shooting range 45, and includes a background region (including a passage, a product shelf, etc.) and a human flow ( 310). K1 to K3 and the like indicate partition areas (passages and the like) in the store 40. A broken line shows the boundary line of a division area. E1 to E4 and the like indicate product shelves and the like. P2 shows a product example. The flow line (trajectory) indicates a change in the position of the person in time series (frame time or the like). A triangular mark indicated by 311 on the flow line 310 of a person indicates an example of the current position (coordinates) of the person.

  A frame region 302 indicates a two-dimensional region associated with the three-dimensional frame region 301. The arrangement configuration in the store 40 is looked down on from above, and information such as a flow line is superimposed. Depending on the installation of the monitoring camera 30 and the setting of the shooting range 45, the detection target can be made two-dimensional in this way. For example, the shooting range 45 may be set in a direction in which the product shelf is viewed from the front. Further, a predetermined association / conversion process may be performed between the three-dimensional information and the two-dimensional information. By making it two-dimensional, for example, it is possible to make information easy for a store manager or the like to see when outputting.

  (S20) Next, the stay detection unit 12 performs a process of calculating the staying degree for each unit region in the imaging range 45. In the present embodiment, a block (B) defined by the matrix (62) is defined as a unit area (divided area), and the staying degree is calculated as a score (s) for each block B. The stay detection unit 12 records / updates the score s for each element (block B) using the data up to S10 and the matrix data 62. The matrix (62) has a matrix / table format. Each element (block B) has information indicating a position in the frame area.

  (S21) The stay detection unit 12 uses the coordinates of the flow line (310, etc.) for each person based on the captured video (person / flow line information 61) as an element (block B) on the matrix (62) corresponding to the frame area. Record the score for each element by plotting (mapping) against. Map appropriately according to the format of the flow line (coordinates). In brief, the score of the element (B) where the flow lines overlap is increased.

  In the matrix 401 (corresponding frame region) in FIG. 4, one frame region, for example, 480 pixels in the vertical direction × 640 pixels in the horizontal direction is divided into a plurality of regions (block B) in the vertical and horizontal directions to form a matrix. In this example, it is divided into 8 parts vertically and horizontally, and 8 × 8 = 64 elements (block B). One element (block B) is a case of 60 × 80 × 480 pixels. Each element (block B) has a score s regarding the staying degree. The score s has an initial value of 0 and is counted up (+1) according to the staying unit (described in S22).

  The matrix 402 is an example in which a person's flow line (for example, 420) is plotted (mapped) on the matrix 401 (element) so that the element score s is counted up (+1). The stay detection unit 12 determines an element (block B) where the flow line 420 (similar to the flow line 310 in FIG. 3) overlaps on the matrix 401, and the score s of the overlap element (B) is determined according to the stay unit. Count up (+1). In the example of 402, the score s of each element (B) through which the flow line 420 passes is an increase such as +1 for an element of one stay unit, +2 for an element of a stay unit, and +5 for an element of a stay unit. is there.

  (S22) The stay detection unit 12 updates the score s of the element (block B) corresponding to the current position of the person in the flow line by incrementing (+1) for each stay unit determination (for example, elapse of stay time). To do. The counting process of S22 is performed over a plurality of frames (time) to be detected and a plurality of flow lines. In the present embodiment, a predetermined frame time unit (e.g., every frame, every second, etc.) and a person unit are considered as the staying unit, and the unit is counted up (+1). For example, when a person continues to stay on the same block B in a plurality of frames, the score s of the element (B) is increased by the amount of the plurality of frames. When a plurality of persons stay on or pass through the same block B, the score s of the element (B) is increased by the number of persons. The staying unit (counting method) is variable depending on the setting.

  The division configuration of the element (block B) of the matrix (62) is variable according to the setting or the like, and the accuracy (effect) of the stay detection can be adjusted according to this division configuration. For example, when the number of elements is reduced (rough), the calculation of the staying location for each frame can be realized at high speed. When the number of elements is increased (finely), the staying location can be detected in finer units. In addition, for example, the division configuration of the element (block B) may be set in association with the configuration such as the product arrangement in the store 40 in the shooting range 45.

  (S30) Next, the stay detection unit 12 uses the processing results up to S20 (matrix data 62 or the like) to extract an element (block B) having a high stay degree (score s) in the matrix (62). Do.

  (S31) In the present embodiment, first, the score s of each element (B) of the matrix (62) is compared with a predetermined threshold value (st). If the score s is greater than or equal to the threshold value st (s ≧ st), the element (block B) is extracted. The threshold value st is variable depending on the setting.

  A matrix 501 (corresponding frame region) in FIG. 5 is an example in which an element (block B) having a score s equal to or greater than a predetermined threshold value st (for example, st = 8) is extracted from the matrix 501 of the processing result example of S22.

  Further, in the matrix 502, as in the case of 501, a plurality of elements (block B) indicated by asterisks are extracted, and the individual elements (block B) extracted in S31 are used as individual “retention elements”. (Standing block) "(first detection result information). The first detection result information is stored as the staying information 63.

  (S32) In addition, the stay detection unit 12 may perform the following process when a plurality of extraction elements (block B) are present in the frame region using the result of S31. The stay detection unit 12 classifies the plurality of extraction elements (B) into one or more cluster sets by clustering. As this clustering method, for example, a known technique such as a K-average method can be used.

  In the matrix 502 of FIG. 5, 520 indicated by each ellipse indicates an example of a cluster set that is a classification result by the clustering. The cluster set information, which is the classification result, can be second detection result information indicating the stay location (stored as stay information 63).

  In particular, when a large number of elements are extracted in S31, classification in S32 has an effect of making it easier to determine the staying location in the frame area.

  (S33) Further, the stay detection unit 12 creates and associates information such as a circle indicating a staying location with respect to the first result information (stay element) in S31 or the second result information (cluster set) in S32. May be. This information can be used as the third detection result information (stored as staying information 63). For example, the center of the circle can be regarded as a person's staying point, and the size (radius) of the circle can be regarded as a staying range.

  The matrix 601 (corresponding frame area) in FIG. 6 is an example in which a circle indicating a staying place is created and associated with the cluster set 520 (second result information) in FIG. As a method of creating a circle, for example, the center point of the circle is a center point with respect to one or more elements (block B) for each cluster set. Further, for example, the radius of the circle is an average value of the distances of the respective elements from the center of gravity, and is used as the radius. Moreover, not only a circle but a figure such as an ellipse may be used.

  (S40) The stay detection unit 12 collects and stores the detection result information obtained in the processes up to S30 as the stay information 63. The staying information 63 may be not only numerical information but also data in a format that can be displayed on a computer terminal screen of a store manager or the like for output processing. The output unit 13 performs an output process using the staying information 63 in accordance with an operation of a store manager or the like. For example, the output unit 13 performs data transmission of the staying information 63, screen display processing, and the like to a computer terminal such as an administrator of the corresponding store 40.

  A frame 602 in FIG. 6 is an output example corresponding to the third detection result information 601, and a circle 610 indicating a staying location of 601 on a background region (such as an image in the absence of a person) of the imaging range 45. Is displayed in a superimposed manner. Similarly, the first and second detection result information may be superimposed and output. In addition, related information such as flow lines may be appropriately superimposed and output. Creating and displaying information such as a circle indicating the staying location has an effect of making it easier to determine the staying location in the frame area.

  By referring to the result information (the stay information 63), the manager of the store 40 can determine and estimate a specific stay location in the area of the imaging range 45. The detected stay location is associated with a specific passage (section), product shelf, product, etc., depending on the imaging range 45, matrix configuration, and the like.

  Still further, the stay detection system 10 uses the stay information 63 and the person / flow line information 61 described above to calculate the number of people who stop by for each stay location and every time zone based on the person discrimination by the person detection function. , May be included in the detection result information. For example, the number of persons may be determined and counted in units such as blocks. Further, for example, as in the example of FIG. 3, the number of persons may be determined and counted in units of sections (boundary lines) set in the area of the imaging range 45. For example, it is possible to calculate the number of people who stop by by determining the passage (in / out) of the boundary line by a person by geometric calculation based on the coordinates on the image and counting the number of persons staying in the section. Thereby, for example, the number of customers visiting the store in front of a specific product shelf can be grasped.

[Effects]
As described above, according to the present embodiment, using the moving image captured by the surveillance camera 30, the retention of a person in the region of the imaging range 45 is quantified as the retention degree (score s), and the retention location is determined. It can be detected and can be used for business management of the store 40 or the like. For example, a stay location (a specific section, a product shelf, a product, or the like) can be detected in a form associated with the arrangement configuration of the product or the like in the store 40. The detected staying place is a place where a person is stationary or staying for a long time, or a place where a plurality of persons gather. Therefore, for example, it can be estimated that it is a place where a visitor is worried and stopped, or a place where a product is searched or picked up.

  The staying information 63 can be used for business management of the store 40. For example, it can be used as management know-how information or in cooperation with a POS system. For example, it is useful for examination of product display and advertisement arrangement in the store 40, and can be used for purchase of products in comparison with product inventory / sales data.

  Other embodiments include the following.

  (1) Depending on the analysis / detection function provided in the monitoring camera server 20 or the like, it may be impossible to distinguish / specify a plurality of persons in a moving image (frame area) (when it is impossible to grasp the flow line for each person). Conceivable. Even in such a case, it is possible to roughly determine the staying degree (staying degree in an unspecified number of units) and detect the staying place. For example, even if there is no flow line information, if the overlap of the object with respect to the frame area (background area, etc.) and the corresponding matrix element (block B) can be detected (for example, the difference in pixel information between the background area and the object is determined). A known method or the like may be used), and the approximate residence degree and residence location can be calculated.

  (2) Depending on the function provided in the monitoring camera server 20 or the like, the information on the current position of the person may have the direction information in consideration of the detected face direction, the direction of the line of sight, and the like. For example, 311 in FIG. 3 shows a case where the product shelf E2 (product P2) is facing. You may utilize said direction information for determination of a staying degree and a staying location. For example, from the element corresponding to the person's current position (block B), calculation is performed so as to increase the staying degree (score s) of the element (block B) in the direction corresponding to the direction information with a predetermined weight.

  (3) The concept of staying (staying degree) described above includes not only a case where one person is stationary, but also a case where a plurality of persons pass through each other. For example, in the case where there are a first section and a second section in the store 40, one store visitor stays long in the first section, and a plurality of store visitors stay in the second section in a short time. The case where it passed is considered. Therefore, it is possible to make a determination by distinguishing both. The first method that distinguishes people by the person / flow line detection function, calculates the staying degree for each person and detects the staying location, and calculates the staying degree by the cumulative unit of multiple people and detects the staying location The second method (the embodiment described above) is possible. When the first method is used, for example, it is determined that the staying degree in the first section is high and the staying degree in the second section is low. When the second method is used, for example, it is determined that both the staying degree in the first section and the staying degree in the second section are high.

  As mentioned above, the invention made by the present inventor has been specifically described based on the embodiment. However, the present invention is not limited to the embodiment, and various modifications can be made without departing from the scope of the invention. Needless to say.

  INDUSTRIAL APPLICABILITY The present invention can be used for a surveillance camera system, in particular, a fixed camera system for the purpose of store management, a system for analyzing residence (congestion) in public transportation, and the like.

DESCRIPTION OF SYMBOLS 10 ... Residence detection system, 11 ... Data acquisition part (frame processing part), 12 ... Residence detection part, 13 ... Output part,
20 ... surveillance camera server, 21 ... surveillance camera control unit, 22 ... image analysis unit, 23 ... moving image data, 24 ... analysis information,
30 ... surveillance camera,
40 ... store, 45 ... shooting range,
60 ... frame data, 61 ... person / flow line information, 62 ... matrix data, 63 ... detection result (staying information), 102 ... person / flow line information.

Claims (9)

Built-in or connected to a fixed camera system that includes a fixed camera that captures a predetermined imaging range, and is a system that detects retention of an object in the area of the imaging range,
A data acquisition unit that performs processing for acquiring information on the position of the object in a frame region corresponding to the imaging range by analysis processing of a moving image to be detected captured by the fixed camera;
Using matrix data having a plurality of elements obtained by dividing a frame region corresponding to the shooting range into a plurality of blocks,
A process of mapping the position of the object in the frame region to a block on the matrix and recording a score for each element corresponding to a block in which the position of the object overlaps;
A process of extracting elements having a high score in the matrix based on a predetermined threshold;
A stay detection system comprising: a stay detection unit that performs processing for outputting information on a block corresponding to an element having a high score as stay result detection result information. Built-in or connected to a fixed camera system that includes a fixed camera that captures a predetermined shooting range, and is a system that detects stagnation of a person in the shooting range area,
A frame region corresponding to the imaging range is acquired based on the frame data or the image analysis information based on the frame data or the image analysis information obtained by acquiring at least one of moving image or still image frame data to be detected and captured by the fixed camera. A data acquisition unit that performs processing for acquiring information on the person or the flow line in
Using matrix data having a plurality of elements obtained by dividing a frame region corresponding to the shooting range into a plurality of blocks,
A process of mapping information on the person or the flow line thereof in the frame region to a block on the matrix and recording a score for each element corresponding to the block on which the person or the flow line overlaps;
A process of extracting elements having a high score in the matrix based on a predetermined threshold;
A stay detection system comprising: a stay detection unit that performs processing for outputting information on a block corresponding to an element having a high score as stay result detection result information. The residence detection system according to claim 2,
In the process of recording the score, the stay detection unit maps the coordinates of the current position of the person in the frame area or the flow line of the person to the block on the matrix, and maps the current position of the person. A dwell detection system characterized in that a process of counting up the score of an element corresponding to the block corresponding to the coordinates of each time is performed every elapse of a predetermined frame time unit. The residence detection system according to claim 2,
In the process of recording the score, the stay detection unit uses the information of the person or its flow line, counts up the score for each element in a cumulative unit of a plurality of persons, and in the process of extracting, A retention detection system, wherein an element having a high cumulative score of the plurality of persons is extracted. The residence detection system according to claim 2,
In the process of recording the score, the stay detection unit counts up the score for each element in units of individual persons using the information of the person or its flow line, and in the process of extracting, the individual A retention detection system characterized by extracting an element having a high score for each person. The residence detection system according to claim 2,
When there are a plurality of elements as information obtained as a result of the extraction process, the stay detection unit performs a process of classifying the cluster set into a cluster set by clustering, and outputs the cluster set information as stay result detection result information. A dwell detection system characterized by that. The residence detection system according to claim 2,
The stay detection unit performs a process of creating and associating information on a circle indicating a stay location with respect to information obtained as a result of the extracting process, and detecting the information on the circle indicating the stay location as a detection result of the stay location. A dwell detection system characterized by outputting as information. The residence detection system according to claim 2,
The staying location corresponds to a specific location in the store by setting the configuration of the spatial arrangement in the store in the shooting range and the configuration of the plurality of blocks of the matrix in association with each other. A dwell detection system characterized by being detected as an attached block. A program that is built in or connected to a fixed camera system including a fixed camera that captures a predetermined imaging range, and that causes the computer to function as a system that detects retention of an object in the area of the imaging range,
Performing a process of acquiring information on a position of an object in a frame region corresponding to the imaging range by an analysis process of a moving image to be detected captured by the fixed camera;
Using matrix data having a plurality of elements obtained by dividing a frame region corresponding to the shooting range into a plurality of blocks,
Mapping the position of the object in the frame region to a block on the matrix and recording a score for each element corresponding to the block where the position of the object overlaps; and
Performing a process of extracting an element having a high score in the matrix based on a predetermined threshold;
And a step of outputting a block corresponding to an element having a high score as detection result information of the staying portion.