JP6866950B2 - Analyzer, control method, and program - Google Patents

Description

The present invention relates to behavioral analysis using image analysis.

Technology for analyzing human behavior using images from surveillance cameras has been developed. Patent Document 1 discloses a technique of deriving a physical quantity representing a human behavior state existing in a monitored area and calculating a person's suspiciousness based on the difference between the physical quantity and the average value. Examples of physical quantities include the time spent in the monitored area.

JP-A-2019-8831

In Patent Document 1, the degree of suspicion of all persons is calculated based on a common average value. However, even normal people who are not suspicious people can make a big difference in their behavior. For example, it is thought that the time required to move to the same place differs greatly between young people and elderly people.

The present invention has been made in view of the above problems, and one of the objects thereof is to provide a technique for performing behavioral analysis using captured images with higher accuracy.

The analyzer of the present invention has 1) a first captured image generated by a first camera installed in the facility and a second captured image generated by a second camera installed in the facility, respectively. From, the detection unit that detects the same person and 2) the movement time required for the person detected by the detection unit to move from the place where the image is captured by the first camera to the place where the person is imaged by the second camera is calculated. Using at least one of the first captured image and the second captured image, a specific unit that specifies the attribute of the person and 4) a reference value corresponding to the specified attribute are acquired and calculated. It has a comparison unit that compares the travel time with the acquired reference value and outputs output information based on the comparison result.

The control method of the present invention is executed by a computer. The control method is as follows: 1) From each of the first captured image generated by the first camera installed in the facility and the second captured image generated by the second camera installed in the facility. Calculation to calculate the movement time required for the detection step to detect the same person and 2) the person detected by the detection step to move from the place imaged by the first camera to the place imaged by the second camera. Using at least one of the step, 3) the first captured image and the second captured image, a specific step for specifying the attribute of a person, and 4) a reference value corresponding to the specified attribute are acquired and calculated, and the travel time is calculated. It has a comparison step of comparing with the acquired reference value and outputting output information based on the comparison result.

The program of the present invention causes a computer to execute each step of the control method of the present invention.

According to the present invention, there is provided a technique for performing behavioral analysis using captured images with higher accuracy.

It is a figure which illustrates the outline of the operation of the analysis apparatus of Embodiment 1. It is a figure which illustrates the structure of the analysis apparatus of Embodiment 1. It is a figure which illustrates the computer for realizing the analysis apparatus. It is a flowchart which illustrates the flow of the process executed by the analysis apparatus of Embodiment 1. It is a block diagram which illustrates the functional structure of the analysis apparatus which has an update part. It is a figure which illustrates the output information. It is a figure which illustrates the warning screen including the warning level further.

Hereinafter, embodiments of the present invention will be described with reference to the drawings. In all drawings, similar components are designated by the same reference numerals, and description thereof will be omitted as appropriate. Further, unless otherwise specified, in each block diagram, each block represents a configuration of a functional unit, not a configuration of a hardware unit.

<Overview>
FIG. 1 is a diagram illustrating an outline of the operation of the analysis device 2000 of the present embodiment. FIG. 1 is a diagram showing a conceptual explanation for facilitating an understanding of the operation of the analysis device 2000, and does not specifically limit the operation of the analysis device 2000.

The analysis device 2000 analyzes the captured image obtained from the camera installed in the facility 40. The facility 40 is provided with at least two cameras (first camera 10 and second camera 20). The first camera 10 and the second camera 20 are installed at different locations in the same facility. The term "inside the facility" is not necessarily limited to one building. For example, the facility 40 may include a plurality of buildings and outdoor places (gardens, etc.) in the vicinity thereof.

Each of the first camera 10 and the second camera 20 repeatedly performs imaging to generate a plurality of captured images. The captured image generated by the first camera 10 is referred to as a first captured image 12, and the captured image generated by the second camera 20 is referred to as a second captured image 22. As the first camera 10 and the second camera 20, for example, a surveillance camera installed in the facility 40 can be used. However, the first camera 10 and the second camera 20 may be cameras provided separately from the surveillance camera.

The analysis device 2000 detects the same person from the first captured image 12 and the second captured image 22. A person detected from both the first captured image 12 and the second captured image 22 is referred to as an analysis target person. The analysis device 2000 is between a place where the person to be analyzed is imaged by the first camera 10 (the image area of the first camera 10) and the place where the person is imaged by the second camera 20 (the image area of the second camera 20). Calculate the time (travel time) required for the person to move.

Further, the analysis device 2000 identifies the attribute of the analysis target person by using at least one of the first captured image 12 and the second captured image 22 in which the analysis target person is detected. As the attributes of the person, various attributes such as age, gender, presence / absence and number of companions can be adopted.

The analysis device 2000 acquires a reference value of the travel time corresponding to the specified attribute. Hereinafter, the reference value of the travel time is also referred to as a reference time. The correspondence between the attribute and the reference time is prepared in advance. The information indicating this association is referred to as reference information 100. The reference information 100 indicates the reference time 104 as the reference value of the movement time of the person having the attribute 102 in association with the attribute 102. For example, the reference information 100 is stored in a storage device accessible from the analysis device 2000.

The analysis device 2000 compares the reference time corresponding to the attribute of the analysis target person with the travel time calculated for the analysis target person, and generates output information regarding the comparison result. For example, the analysis device 2000 determines whether or not the travel time is equal to or longer than the reference time, and generates output information indicating a warning when the travel time is equal to or longer than the reference time. However, the method of comparing the travel time and the reference time and the output information based on the comparison result are not limited to the examples described here.

For example, in FIG. 1, the analysis device 2000 detects the same person from the first captured image 12 generated at time t1 and the second captured image 22 generated at time t2, and treats this person as the person to be analyzed. The analysis device 2000 calculates t2-t1 as the movement time of the person to be analyzed. Further, the analysis device 2000 identifies a1 as an attribute of the person to be analyzed, and acquires r1 which is the reference time 104 associated with the attribute 102 of a1 in the reference information 100. Then, the analysis device 2000 generates output information by comparing the travel time t2-t1 with the reference time r1.

<Typical action>
The time when a person moves in the facility can be used as various indicators. For example, from the viewpoint of security, it is possible to detect a person who travels too long as a person requiring attention. In addition, the evaluation of the length of travel time can be realized by comparison with the reference time. For example, a method such as "if the travel time is longer than the reference time, it is determined that the travel time is too long" can be considered.

However, if a value common to all people is used as the reference time, it is difficult to properly evaluate the length of travel time. This is because the appropriateness of a person's travel time depends on the person's attributes. For example, it can be said that the travel time of an elderly person is generally longer than the travel time of a young person. Therefore, if the same reference time is used to evaluate the travel time of the elderly and the travel time of the young, it is difficult to make an appropriate evaluation.

According to the analysis device 2000 of the present embodiment, the movement time of the person to be analyzed is compared with the reference time determined according to the attribute of the person. Therefore, the travel time can be evaluated more appropriately than in the case where the travel time of the analysis person is compared with the reference time common to all objects. That is, it is possible to analyze human behavior by analyzing captured images with higher accuracy.

Hereinafter, the analysis device 2000 of the present embodiment will be described in more detail.

<Example of functional configuration of analyzer 2000>
FIG. 2 is a diagram illustrating the configuration of the analysis device 2000 of the first embodiment. The analysis device 2000 includes a detection unit 2020, a calculation unit 2040, a specific unit 2060, and a comparison unit 2080. The detection unit 2020 detects the same person from the first captured image 12 and the second captured image 22. A person detected from both the first captured image 12 and the second captured image 22 is treated as a person to be analyzed. The calculation unit 2040 calculates the movement time required for the person to be analyzed to move between the image pickup area of the first camera 10 and the image pickup area of the second camera 20. The identification unit 2060 identifies the attributes of the person to be analyzed by using at least one of the first captured image 12 and the second captured image 22. The comparison unit 2080 acquires the reference time 104 corresponding to the attribute of the person to be analyzed from the reference information 100. The comparison unit 2080 compares the movement time of the person to be analyzed with the acquired reference time 104, and generates output information based on the comparison result. The comparison unit 2080 outputs the output information.

<Hardware configuration of analyzer 2000>
Each functional component of the analyzer 2000 may be realized by hardware (eg, a hard-wired electronic circuit) that realizes each functional component, or a combination of hardware and software (eg, electronic). It may be realized by a combination of a circuit and a program that controls it). Hereinafter, a case where each functional component of the analysis device 2000 is realized by a combination of hardware and software will be further described.

FIG. 3 is a diagram illustrating a computer 1000 for realizing the analysis device 2000. The computer 1000 is an arbitrary computer. For example, the computer 1000 is a personal computer (PC), a server machine, a tablet terminal, a smartphone, or the like. The computer 1000 may be a dedicated computer designed to realize the analysis device 2000, or may be a general-purpose computer.

The computer 1000 includes a bus 1020, a processor 1040, a memory 1060, a storage device 1080, an input / output interface 1100, and a network interface 1120. The bus 1020 is a data transmission line for the processor 1040, the memory 1060, the storage device 1080, the input / output interface 1100, and the network interface 1120 to transmit and receive data to and from each other. However, the method of connecting the processors 1040 and the like to each other is not limited to the bus connection. The processor 1040 is a processor such as a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), or an FPGA (Field-Programmable Gate Array). The memory 1060 is a main storage device realized by using RAM (Random Access Memory) or the like. The storage device 1080 is an auxiliary storage device realized by using a hard disk drive, an SSD (Solid State Drive), a memory card, a ROM (Read Only Memory), or the like. However, the storage device 1080 may be composed of hardware similar to the hardware constituting the main storage device, such as RAM.

The input / output interface 1100 is an interface for connecting the computer 1000 and the input / output device. The network interface 1120 is an interface for connecting the computer 1000 to the communication network. This communication network is, for example, LAN (Local Area Network) or WAN (Wide Area Network). The method of connecting the network interface 1120 to the communication network may be a wireless connection or a wired connection. For example, the analysis device 2000 is communicably connected to the first camera 10 and the second camera 20 via the network interface 1120.

The storage device 1080 stores a program module that realizes a functional component of the analysis device 2000. The processor 1040 realizes the function corresponding to each program module by reading each of these program modules into the memory 1060 and executing the program module.

For example, the reference information 100 is stored in the storage device 1080. However, the reference information 100 may be stored in a storage device provided outside the analysis device 2000.

<About the first camera 10 and the second camera 20>
Each of the first camera 10 and the second camera 20 has a function of performing imaging and generating image data representing the result. For example, the first camera 10 and the second camera 20 generate moving image data, respectively. In this case, the first captured image 12 and the second captured image 22 are moving image frames constituting the moving image data, respectively. However, the first camera 10 and the second camera 20 may be still cameras configured to repeatedly generate still images.

The first camera 10 and the second camera 20 may execute a part of the process described as the process executed by the analysis device 2000. For example, a process of detecting a person from the first captured image 12 (for example, a process of generating detection information described later) may be executed by the first camera 10. In this case, the first camera 10 transmits information about the detected person (for example, detection information described later) to the analysis device 2000. By receiving this information, the detection unit 2020 grasps the information about the person detected from the first captured image 12.

The same applies to the second camera 20. That is, the second camera 20 may detect a person from the second captured image 22 and transmit information about the detected person to the analysis device 2000. By receiving this information, the analysis device 2000 grasps the information about the person detected from the second captured image 22.

<Example of usage scene>
The analysis device 2000 can be used in various scenes. For example, the analyzer 2000 is used for person monitoring in various facilities such as airports and shopping malls. For example, when the passage is to be monitored, the first camera 10 and the second camera 20 are installed at the entrance and the exit, respectively. By doing this, it is possible to monitor the person passing through this passage.

For example, the comparison unit 2080 determines whether or not the movement time of the analysis target person is longer than the reference time. If the travel time is longer than the reference time, it is possible that the person has taken some suspicious behavior in the middle of the passage because it took too long to move through the passage.

Here, if there is a place in the aisle that requires consideration of privacy, such as a changing space or a toilet, it is difficult to directly monitor the place with a surveillance camera. Therefore, by comparing the time required for moving the passage with the reference time, it is possible to indirectly grasp whether or not the person to be analyzed has behaved suspiciously.

<Processing flow>
FIG. 4 is a flowchart illustrating the flow of processing executed by the analysis device 2000 of the first embodiment. The detection unit 2020 detects the same person from the first captured image 12 and the second captured image 22 (S102). The calculation unit 2040 calculates the travel time of the person to be analyzed (S104). The identification unit 2060 specifies the attributes of the person to be analyzed (S106). The comparison unit 2080 acquires the reference time 104 corresponding to the attribute of the person to be analyzed from the reference information 100 (S108). The comparison unit 2080 compares the calculated travel time with the acquired reference time 104 to generate output information (S110). The comparison unit 2080 outputs the output information (S112).

<Acquisition method of the first captured image 12>
The detection unit 2020 acquires the first captured image 12. Here, an existing technique can be used as a technique for acquiring an image captured by the camera. For example, the first camera 10 transmits the first captured image 12 to the analysis device 2000 each time a new first captured image 12 is generated. Then, the detection unit 2020 acquires a new first captured image 12 by receiving the first captured image 12 transmitted from the first camera 10. In addition, for example, the first camera 10 may store the first captured image 12 in a storage device accessible from the analysis device 2000. In this case, for example, the detection unit 2020 periodically accesses the storage device and acquires the first captured image 12 actually acquired.

<Acquisition method of the second captured image 22>
The method of acquiring the second captured image 22 is the same as the method of acquiring the first captured image 12.

<Method of detecting the same person from the first captured image 12 and the second captured image 22: S102>
The detection unit 2020 detects the same person from the first captured image 12 and the second captured image 22 (S102). For example, the detection unit 2020 detects a person from the first captured image 12 by executing an object detection process on the first captured image 12. Then, the detection unit 2020 generates information (hereinafter, detection information) indicating the association of the detected person as "identifier of the first camera 10, person identifier, detection time, image feature of the person".

Here, when the first camera 10 takes images a plurality of times while the person passes through the imaging region of the first camera 10, the same person can be repeatedly detected from the plurality of first captured images 12. Therefore, the detection unit 2020 updates the detection information about the same person while the same person continues to be detected from the first captured image 12. Specifically, the detection information is updated by adding the generation time of the first captured image 12 in which the person is detected and the image features detected from the first captured image 12 to the detection information.

Similarly, the detection unit 2020 also generates detection information for the person detected from the second captured image 22.

The detection unit 2020 records information on a person located in an area between the first camera 10 and the second camera 20 (hereinafter, a target area). This information is called candidate information. The candidate information includes detection information for each person located in the target area. The camera identifier indicated by the detection information in the candidate information is information indicating whether the person entered the target area through the imaging region of the first camera 10 or the imaging region of the second camera 20.

For example, when the detection unit 2020 generates detection information for a person detected from the first captured image 12, it determines whether or not the detection information for the same person as that person is included in the candidate information. This determination can be realized by comparing the image feature indicated by the detection information generated by using the first captured image 12 with the image feature indicated by each detection information included in the candidate information. If the candidate information does not include the detection information about the same person as the person detected from the first captured image 12, the detection unit 2020 uses the detection information about the person detected from the first captured image 12 as the candidate information. Add. By doing so, it is possible to record that the person has newly entered the target area through the imaging area of the first camera 10.

It is assumed that the candidate information includes the detection information about the same person as the person detected from the first captured image 12. In this case, the detection unit 2020 determines which of the first camera 10 and the second camera 20 is the identifier of the camera indicated by the detection information. When this identifier is the identifier of the second camera 20, it means that the person detected from the first captured image 12 has moved from the imaging region of the second camera 20 to the imaging region of the first camera 10. That is, it means that the same person is detected from both the first captured image 12 and the second captured image 22. Therefore, the detection unit 2020 identifies the person as the person to be analyzed. Since the person has gone out of the target area, the detection unit 2020 deletes the detection information of the person from the candidate information.

On the other hand, it is assumed that the candidate information includes the detection information about the same person as the person detected from the first captured image 12, but the identifier of the camera indicated by the detection information is the identifier of the first camera 10. .. In this case, the person enters the target area through the image pickup area of the first camera 10, and then passes through the image pickup area of the first camera 10 again without passing through the image pickup area of the second camera 20. It means that it came out of. Therefore, the travel time cannot be calculated for this person. Therefore, the detection unit 2020 does not specify the person as the analysis target person, but deletes the detection information of the person from the candidate information.

The same process is performed when a person is detected from the second captured image 22. Specifically, the detection unit 2020 includes detection information of the same person as the person detected from the second captured image 22 in the candidate information, and the identifier of the camera shown in the detection information is the first. If it is the identifier of the camera 10, the detection unit 2020 identifies the person as the person to be analyzed. In this case, this is because it means that the person entered the target area through the image pickup area of the first camera 10 and then exited the target area through the image pickup area of the second camera 20. On the other hand, if the candidate information includes the detection information of the same person as the person detected from the second captured image 22, but the identifier of the camera shown in the detection information is the identifier of the second camera 20. The detection unit 2020 deletes the detection information of the person from the candidate information without specifying the person as the person to be analyzed. Further, if the candidate information does not include the detection information of the same person as the person detected from the second captured image 22, the detection unit 2020 uses the detection information generated by using the second captured image 22 as the candidate information. to add.

<Calculation method of travel time: S104>
The calculation unit 2040 calculates the travel time of the person to be analyzed (S104). For example, the movement time of the person to be analyzed is the difference between the time when the person to be analyzed is detected from the imaging range of the first camera 10 and the time when the person to be analyzed is detected from the imaging range of the second camera 20 (the later one). It can be calculated as the value obtained by subtracting the earlier time from the time). The time when the person to be analyzed is detected in the imaging range of the first camera 10 is shown in the detection information generated by using the first captured image 12. Similarly, the time when the person to be analyzed is detected in the imaging range of the second camera 20 is shown in the detection information generated by using the second captured image 22.

Here, in many cases, the person to be analyzed is included in the plurality of first captured images 12 and the plurality of second captured images 22. This is because it is considered that the camera captures images a plurality of times while the person to be analyzed passes through the imaging range of the camera. Therefore, the detection information includes a plurality of detection times.

It is assumed that the detection information generated by using the first captured image 12 includes a plurality of detection times. In this case, the calculation unit 2040 determines the time when the person to be analyzed has passed the imaging range of the first camera 10 based on the plurality of detection times. For example, the calculation unit 2040 treats the latest detection time, the earliest detection time, or the average value thereof as the time when the person to be analyzed has passed the imaging range of the first camera 10. The same applies to the method of determining the time when the person to be analyzed has passed the imaging range of the second camera 20 by using the detection information generated by using the second captured image 22.

<Specification of attributes: S106>
The identification unit 2060 specifies the attributes of the person to be analyzed (S106). Various attributes such as age group, gender, presence / absence and number of companions can be adopted. It should be noted that existing techniques can be used for techniques for identifying various attributes of a person by analyzing an image containing the person.

Here, both the first captured image 12 and the second captured image 22 may be used, or only one of them may be used to specify the attributes of the person to be analyzed. Further, when a plurality of first captured images 12 include a person to be analyzed, the identification unit 2060 may specify the attributes by using only a part (for example, one) of the plurality of first captured images 12. Alternatively, the attributes may be specified by using all the first captured images 12. The same applies to the case where the second captured image 22 is used.

<About standard information 100>
The reference information 100 indicates the reference time 104 in association with the attribute 102. Here, the reference time 104 associated with the attribute 102 may be one time or a time range. In the latter case, for example, a time range such as "10 minutes or more and 15 minutes or less" is defined as the reference time 104.

Further, a plurality of reference times 104 may be associated with one attribute 102. For example, three reference times 104 of "10 minutes, 20 minutes, and 30 minutes" are associated with one attribute 102. In this case, for example, the comparison unit 2080 determines whether the travel time corresponds to "less than 10 minutes", "10 minutes or more and less than 20 minutes", "20 minutes or more and less than 30 minutes", or "30 minutes or more". To do.

Various methods can be adopted as the method for determining the reference time 104 corresponding to the attribute 102. For example, the actual value of the travel time is accumulated for each attribute, and the reference time corresponding to the attribute is set as the statistical value (average value, etc.) of the actual value. For example, a test period is provided before the operation of the analysis device 2000, and the analysis device 2000 is operated for the facility 40 during the test period, and the travel time is calculated and the attributes are specified for each analysis target person. Then, the travel time is accumulated as an actual value corresponding to the attribute. By doing so, it is possible to obtain a plurality of actual values of the travel time at the facility 40 for each attribute. The analysis device 2000 calculates the statistical value of the actual value for each attribute, determines the reference time corresponding to the attribute based on the statistical value, and generates the reference information 100 in which the attribute and the reference time are associated with each other.

For example, the reference time corresponding to an attribute is determined as a statistical value of the actual value obtained for that attribute. In addition, for example, the reference time corresponding to the attribute may be a value obtained by adding or subtracting a predetermined margin from the statistical value of the actual value obtained for the attribute. For example, the comparison unit 2080 determines whether or not the travel time is equal to or longer than the reference time, and outputs output information indicating a warning when the travel time is equal to or longer than the reference time. In this case, it is preferable to set a value obtained by adding a predetermined margin to the statistical value of the actual value as the reference time. By doing so, it is possible to allow the travel time to be slightly longer than the actual statistical value.

Here, as described above, a plurality of reference times 104 may be associated with one attribute. In this case, for example, the analysis device 2000 calculates the distribution of the travel time obtained for the attribute, divides the calculated distribution into a plurality of ranges, and sets the boundary value of each range as the reference time 104. As a method of dividing the distribution of the travel time, for example, a method of dividing the distribution of the travel time into a predetermined number so that the number of samples in the sample is equal can be considered.

The analysis device 2000 may update the reference information 100 by using the travel time calculated at the time of operation after the reference information 100 is generated and the operation of the analysis device 2000 is started. The functional component unit that updates the reference information 100 is called an update unit. FIG. 5 is a block diagram illustrating the functional configuration of the analysis device 2000 having the update unit 2100.

The update unit 2100 updates the reference time 104 corresponding to the attribute 102 that matches the attribute specified by the specific unit 2060, using the travel time calculated by the calculation unit 2040. In order to realize this update, for example, the correspondence between the attributes calculated during the test period and the operation and the travel time is recorded in the storage device. The update unit 2100 acquires the usage time calculated so far for the attribute specified by the specific unit 2060 from the storage device. Then, the update unit 2100 calculates a statistical value of each acquired usage time and a usage time newly calculated by the calculation unit 2040, and updates the reference time 104 with the statistical value. However, in the case of a statistical value such as an average value that can be calculated if the number of samples is known, it is not necessary to record all the specific movement times, and the number of samples for each attribute may be recorded.

The method of determining the reference time 104 corresponding to the attribute 102 is not limited to the one using the actual value of the travel time. For example, the reference time 104 may be set manually.

<Acquisition of reference time: S108>
The comparison unit 2080 acquires the reference time corresponding to the attribute specified by the specific unit 2060 (S108). Specifically, the comparison unit 2080 identifies an attribute 102 that matches the attribute of the person to be analyzed from the reference information 100, and acquires the reference time 104 shown in the reference information 100. Here, the attribute 102 that matches the attribute of the person to be analyzed is an attribute 102 that is the same as the attribute of the person to be analyzed, or an attribute 102 that includes the attribute of the person to be analyzed. In the latter case, for example, the attribute 102 indicates the range of the attribute, and the attribute of the person to be analyzed is included in the range.

<Comparison of reference time and travel time and generation of output information: S110>
The comparison unit 2080 compares the reference time and the travel time, and generates output information based on the result (S110). Here, various things can be adopted for comparing the reference time and the travel time. For example, the comparison unit 2080 determines whether or not the travel time is equal to or longer than the reference time. By doing so, it is possible to detect a person who travels for too long.

When the movement time of the person to be analyzed is equal to or longer than the reference time, for example, the comparison unit 2080 generates output information indicating a warning. This output information includes, for example, a warning message and detailed information regarding the detection of the person to be analyzed. The detailed information includes, for example, the detection time and the image of the person to be analyzed.

Further, it is assumed that there are a plurality of places in the facility 40 for which the travel time is calculated. That is, it is assumed that a plurality of sets of the first camera 10 and the second camera 20 are provided in the facility 40. In this case, an identifier is assigned to a place (that is, a set of cameras) for which the travel time is calculated. For example, this identifier is the name assigned to the location. It is preferable to further include this identifier in the detailed information described above.

FIG. 6 is a diagram illustrating output information. In FIG. 6, the output information is the warning screen 50. The warning screen 50 shows the warning message 52 and the detailed information 54. The detailed information 54 includes the name of the detection location, the detection time, and an image of the person to be analyzed.

The information included in the detailed information 54 is not limited to that shown in FIG. For example, the detailed information 54 may further include the attributes of the person to be analyzed. At this time, the detailed information 54 may further include an attribute 102 determined to match the attribute in addition to the attribute of the person to be analyzed. Further, the detailed information 54 may further include a travel time calculated for the person to be analyzed, a reference time compared with the travel time, and the like. This point is the same for the example of FIG. 7 described later.

Here, when the travel time is less than the reference time, the comparison unit 2080 may or may not output the output information. By outputting the output information even when the travel time is less than the reference time, for example, all the comparison results for each analysis target person can be recorded.

The comparison between the reference time and the travel time is not limited to "whether or not the travel time is longer than the reference time". For example, the comparison unit 2080 may determine whether or not the travel time is equal to or less than the reference time. When the travel time is equal to or less than the reference time, for example, the comparison unit 2080 outputs the same output information as the warning screen described above. This is because if the travel time is too short, there is a possibility that a situation should be warned, for example, a transportation means prohibited from being used in the facility 40 is used.

Further, it is assumed that the reference time indicates a time range. In this case, the comparison unit 2080 determines whether or not the travel time is included in the time range indicated by the reference time. When the travel time is not included in the time range indicated by the reference time, for example, the comparison unit 2080 outputs the same output information as the warning screen described above.

Further, it is assumed that a plurality of reference times 104 are associated with one attribute. In this case, the comparison unit 2080 may compare the travel time with each of the plurality of reference times corresponding to the attributes of the person to be analyzed. For example, the comparison unit 2080 uses the acquired plurality of reference times in order from the shortest to determine "whether or not the travel time is equal to or longer than the reference time". When the travel time is longer than any of the reference times, the comparison unit 2080 is the minimum that satisfies the reference time first determined as "the travel time is equal to or longer than the reference time" (that is, "travel time ≥ reference time"). Reference time) is specified. If the travel time is shorter than the minimum reference time, the comparison unit 2080 determines that the travel time is shorter than any reference time.

When performing such a comparison, the comparison unit 2080 outputs output information based on the comparison with a plurality of reference times. For example, the warning level is associated with the reference time. The height of the warning level is set so that the larger the minimum reference time that satisfies "travel time ≥ reference time", the higher the warning level. By doing so, it is possible to grasp the degree of warning (how suspicious) for a person whose travel time is longer than the reference time.

For example, the correspondence between the reference time and the warning level is defined as "10 minutes: warning level 1", "20 minutes: warning level 2", and "30 minutes: warning level 3". In this case, if the travel time is 25 minutes, the minimum reference value that satisfies the travel time ≥ reference value is 20 minutes. Therefore, the warning level of the person to be analyzed is 2. FIG. 7 is a diagram illustrating a warning screen further including a warning level. The association between the reference time and the warning level is included in the reference information 100 in advance.

The mode of the output information is not limited to the screen described above. For example, the output information may be text information or information converted into a predetermined table format for recording in a database.

<< Correction of reference time >>
Here, the speed at which a person moves is expected to differ depending on the degree of congestion (density of people) at the place where the person moves. Therefore, the comparison unit 2080 may make a correction based on the degree of congestion of the moving place in the reference time instead of using the reference time as it is. For example, the comparison unit 2080 calculates the degree of congestion between the imaging range of the first camera 10 and the imaging range of the second camera 20, and multiplies the correction coefficient based on the degree of congestion by the reference time to obtain the reference time. To correct. The correction coefficient is a real number of 1 or more. Then, the comparison unit 2080 compares the corrected reference time with the travel time.

For example, the degree of congestion can be determined by the number of people located between the imaging region of the first camera 10 and the imaging region of the second camera 20. The number of people located between the imaging region of the first camera 10 and the imaging region of the second camera 20 can be grasped as the number of detection information included in the candidate information described above. In addition, for example, another camera is provided so as to take a bird's-eye view between the first camera 10 and the second camera 20, and the number of people is specified by analyzing the captured image obtained from the camera. You may.

Various methods can be adopted for calculating the correction coefficient. For example, a reference congestion degree (hereinafter referred to as a reference congestion degree) is determined in advance. Then, the comparison unit 2080 uses a value obtained by dividing the degree of congestion when the person to be analyzed is detected by the reference degree of congestion as a correction coefficient. In addition, for example, a function for converting the degree of congestion into a correction coefficient may be defined, and this function may be used to calculate the correction coefficient. Any monotonous non-decreasing function that outputs a value of 1 or more can be used for this function.

When making corrections according to the degree of congestion, it is preferable to consider the degree of congestion when generating the reference information 100 using the actual value of the travel time. For example, the correction coefficient described above is calculated when accumulating the actual value of the travel time. Then, the reference time is calculated using the statistical value obtained by multiplying the actual value of the travel time by the reciprocal of the correction coefficient. By doing so, it is possible to determine the reference time after changing the actual value of the travel time to the predicted value at the reference congestion degree. That is, the reference time can be set so as not to be affected by the degree of congestion when the actual value is calculated.

<Output of output information: S112>
The comparison unit 2080 outputs the output information (S112). The output information can be output by any method. For example, the comparison unit 2080 displays on a display device connected to the analysis device 2000. In addition, for example, the comparison unit 2080 stores the output information in a predetermined storage device. In addition, for example, the comparison unit 2080 transmits to another device (for example, a mobile terminal owned by a guard). It is preferable that the output information transmitted to the other device is displayed on the display device provided in the device. An existing technique can be used as a technique for displaying information transmitted to another device on a display device provided in the device.

Although the embodiments of the present invention have been described above with reference to the drawings, these are examples of the present invention, and various configurations other than the above can be adopted.

For example, the analysis device 2000 can be used for various purposes other than person monitoring. Marketing is an example of another application. Specifically, the degree of interest of the analysis target person in the facility 40 can be grasped based on the travel time of the analysis target person.

For example, suppose the facility is a shopping mall. In this case, the first camera 10 and the second camera 20 are installed at the entrances and exits of different areas in each of the areas such as the children's goods section and the women's clothing section. By doing so, the analysis device 2000 can grasp the movement time in the area for the person to be analyzed.

Here, if the person to be analyzed moves in a certain area for a long time, it is highly probable that the person is interested in the area. On the other hand, if the person to be analyzed moves in a certain area in a short time, it is highly probable that the person is not interested in the area. Therefore, the degree of interest of the person to be analyzed can be grasped by comparing the travel time with the reference time.

By setting a plurality of reference times 104 corresponding to one attribute 102, the degree of interest of the person to be analyzed may be grasped in the same manner as grasping the warning level. Specifically, the shorter the reference time, the smaller the interest level is associated with the plurality of reference times corresponding to one attribute 102. The analysis device 2000 specifies the minimum reference time that satisfies "travel time ≥ reference time", and specifies the level of interest corresponding to the specified reference time as the degree of interest of the person to be analyzed.

The method of dividing the area is not limited to the set of shops where similar items such as "children's clothing section" are placed. For example, there may be one area for each store. By doing this, the degree of interest of each person can be grasped for each store.

Some or all of the above embodiments may also be described, but not limited to:
1. 1. The same person is detected from each of the first captured image generated by the first camera installed in the facility and the second captured image generated by the second camera installed in the facility. With the detector
A calculation unit that calculates the movement time required for a person detected by the detection unit to move between a place imaged by the first camera and a place imaged by the second camera.
Using at least one of the first captured image and the second captured image, a specific unit that identifies the attribute of the person, and a specific unit.
An analysis device having a comparison unit that acquires a reference value corresponding to the specified attribute, compares the calculated travel time with the acquired reference value, and outputs output information based on the comparison result.
2. The reference value corresponding to the attribute is a statistical value of the travel time required by a plurality of persons having the attribute to move between the place imaged by the first camera and the place imaged by the second camera. Is 1. The analyzer described in.
3. 3. 1. It has an update unit that updates the reference value corresponding to the specified attribute using the calculated travel time. Or 2. The analyzer described in.
4. The comparison unit
The degree of congestion of people between the place imaged by the first camera and the place imaged by the second camera is calculated.
Based on the calculated congestion degree, the acquired reference value is corrected based on the calculated congestion degree.
1. Compare the corrected reference value with the calculated travel time. To 3. The analyzer according to any one.
5. The comparison unit outputs output information indicating a warning when the calculated movement time is equal to or longer than the acquired reference value. To 4. The analyzer according to any one.
6. A control method performed by a computer
The same person is detected from each of the first captured image generated by the first camera installed in the facility and the second captured image generated by the second camera installed in the facility. Detection step and
A calculation step for calculating the movement time required for a person detected by the detection step to move between a place imaged by the first camera and a place imaged by the second camera, and a calculation step.
A specific step of identifying the attribute of the person using at least one of the first captured image and the second captured image, and
A control method having a comparison step of acquiring a reference value corresponding to the specified attribute, comparing the calculated travel time with the acquired reference value, and outputting output information based on the comparison result.
7. The reference value corresponding to the attribute is a statistical value of the travel time required by a plurality of persons having the attribute to move between the place imaged by the first camera and the place imaged by the second camera. Is, 6. The control method described in.
8. 6. It has an update step of updating the reference value corresponding to the identified attribute using the calculated travel time. Or 7. The control method described in.
9. In the comparison step
The degree of congestion of people between the place imaged by the first camera and the place imaged by the second camera is calculated.
Based on the calculated congestion degree, the acquired reference value is corrected based on the calculated congestion degree.
6. Compare the corrected reference value with the calculated travel time. ~ 8. The control method according to any one.
10. In the comparison step, when the calculated movement time is equal to or longer than the acquired reference value, output information indicating a warning is output. ~ 9. The control method according to any one.
11. 6. To 10. A program that causes a computer to execute each step of the control method described in any one of them.

10 1st camera 12 1st captured image 20 2nd camera 22 2nd captured image 40 Facility 50 Warning screen 52 Warning message 54 Detailed information 100 Reference information 102 Attribute 104 Reference time 1000 Computer 1020 Bus 1040 Processor 1060 Memory 1080 Storage device 1100 Output interface 1120 Network interface 2000 Analyzer 2020 Detection unit 2040 Calculation unit 2060 Specific unit 2080 Comparison unit

Claims (5)

A detection unit that detects the same person from each of the first captured image generated by the first camera and the second captured image generated by the second camera.
A calculation unit that calculates the movement time required for a person detected by the detection unit to move between a place imaged by the first camera and a place imaged by the second camera.
A specific part that identifies the attributes of the person,
An analysis device having the specified attribute and an output unit that outputs output information based on the calculated travel time. The output information further includes statistical values of travel time required by a plurality of persons having the attribute to move between a place imaged by the first camera and a place imaged by the second camera, and the above-mentioned output information. The analysis device according to claim 1, which is based on a comparison result with the calculated travel time. The output unit
The degree of congestion of people between the place imaged by the first camera and the place imaged by the second camera is calculated.
Based on the calculated congestion degree, the acquired statistical value is corrected based on the calculated congestion degree.
The analysis device according to claim 2, wherein the corrected statistical value is compared with the calculated travel time. A control method performed by a computer
A detection step of detecting the same person from each of the first captured image generated by the first camera and the second captured image generated by the second camera.
A calculation step for calculating the movement time required for a person detected by the detection step to move between a place imaged by the first camera and a place imaged by the second camera, and a calculation step.
Specific steps to identify the attributes of the person,
A control method having the specified attribute and an output step for outputting output information based on the calculated travel time. A program that causes a computer to execute each step of the control method according to claim 4.

Images