JP6678512B2 - Sensor installation position support device - Google Patents

Description

  The present invention relates to a technique for determining an arrangement of a plurality of sensors for performing object detection and position measurement.

  In public facilities such as large commercial facilities, train stations, and airports, in addition to conventional security services such as detection of people and suspicious objects, by measuring and tracking the position of people, spatial value such as marketing, congestion reduction, layout improvement, etc. There is a growing demand for enhanced services. These services can be realized by installing a plurality of sensors such as monitoring cameras and laser radars, and referring to each measurement result. However, to maintain stable measurement accuracy, the arrangement of the sensors is important.

  For example, if you want to detect a suspicious person, you need to install a sensor so that blind spots do not occur, and if you want to track a person, you need to determine the position of the sensor so that the trajectory of the person can be measured without interruption . Therefore, when introducing a service, an expert who actually performs the installation work decides the sensor arrangement and setting based on his own know-how, taking into account the needs of the customer.

  However, this method not only leads to an increase in the installation cost, but also the measurement accuracy is affected by differences in the knowledge and experience of specialists. Therefore, a support technology for automatically determining the sensor installation position is required.

  For example, in Patent Literature 1, in a commercial facility, the distribution of cameras and the installation position of the cameras are determined by estimating an area where persons are likely to gather in accordance with information such as POS (Point of Sales) data and gender of a customer. In Patent Document 2, an optimal installation position of a monitoring camera is automatically output according to a customer's measurement needs such as detection of a suspicious person or movement.

JP 2005-286619 A JP 2012-10210A

  Patent Literature 1 has a problem that although the installation position of the camera can be estimated, detailed installation information such as the installation angle of the camera such as a depression angle cannot be output. In addition, the total number of cameras to be used must be determined in advance, and the determination requires expert know-how and the like.

  In Patent Literature 2, a camera installation position can be automatically estimated from a layout of a shooting range and a customer needs table with a configuration that minimizes the number of cameras. However, there is room for further improvement in high-precision measurement depending on the situation in consideration of actual movement of a person or the like in the photographing range or in a case where a sensor other than the surveillance camera is also used.

  An object of the present invention is to solve the above-mentioned problem, and an object of the present invention is to provide a sensor installation position support device that can appropriately determine an installation position of a sensor such as a camera in consideration of a human flow. I do.

In order to achieve the above object, a sensor installation position support device of the present invention provides a measurement model generation unit (for example, a measurement model generation unit) that generates a measurement model indicating a measurement range of the sensor based on characteristic information of the sensor and measurement item information. Unit 3) and a human flow map generating means (for example, a human flow map) for generating a human flow map (for example, 23 in FIG. 4) indicating the human flow direction of the area based on the map information of the area to be monitored and the human flow information of the area. A sensor arrangement for calculating a measurement sensitivity (for example, Equation 1) for each cell constituting the area based on the measurement model and the human flow map, and determining a sensor arrangement position in the area based on the calculated measurement sensitivity. determining means (e.g., a sensor arrangement adjusting section 5) possess the sensor arrangement determining means, by matching the measurement accuracy included in the measurement model human stream map, a total of each cell The sensitivity is calculated, and the measurement model includes the measurement accuracy according to the distance from the sensor and the measurable range. When the ratio of the measurement accuracy to 1 is A, the measurement sensitivity is measured with respect to the cell area. The accuracy is a value obtained by integrating the ratio of the area of A and A. Other aspects of the present invention will be described in embodiments described later.

  ADVANTAGE OF THE INVENTION According to this invention, the installation position of a sensor, such as a camera, can be determined appropriately in consideration of a human flow.

FIG. 2 is a diagram showing functional blocks of the sensor installation position support device according to the first embodiment. FIG. 7 is a diagram illustrating an example of sensor characteristic information acquired by a measurement model generation unit. FIG. 5 is a diagram illustrating an example of a measurement model generated by a measurement model generation unit. It is a figure explaining the processing of the people flow map generation part. It is a flowchart which shows the process of a person flow map generation part. It is a figure showing the functional block of a sensor arrangement adjustment part. It is a figure showing an example of an important measurement point outputted by a measurement center point detection part. 9 is a flowchart illustrating processing of a measurement model matching unit. It is a figure explaining processing S3 of a measurement model collation part, (a) is a person flow map, (b) is a figure showing a measurement direction of a measurement model, and (c) shows an installation position of a camera. It is a figure explaining processing S5 of a measurement model collation part, (a) is a measurement model, (b) is a measurement model collated with people flow map 51, (c) (d) is an enlarged view of a cell in a people flow map. is there. It is a figure explaining processing S4 of a measurement model collation part. FIG. 9 is a diagram illustrating an example of collating a measurement model by a measurement model collation unit. It is a figure showing an example of a sensor position outputted by a sensor arrangement adjustment part. It is a figure showing the functional block diagram of the sensor installation position support device concerning Embodiment 2. 9 is a flowchart illustrating a process of a sensor arrangement adjustment unit according to the second embodiment. It is a figure showing an example of a sensor position adjusted by a sensor arrangement adjustment part.

An embodiment for carrying out the present invention will be described in detail with reference to the drawings as appropriate.
<< Embodiment 1 >>
FIG. 1 is a diagram illustrating functional blocks of a sensor installation position support device 1 according to the first embodiment. In the present embodiment, an example in which a monitoring camera is used as a sensor will be described. However, the present invention is not limited to these sensors, and different types of sensors may be used in combination.

  The sensor installation position support apparatus 1 holds a database 2 including map information 100 of a target area, human flow information 101, sensor information 102, and measurement item information 103, and as a processing unit, a measurement model generation unit 3 (measurement model generation Means), a human flow map generation unit 4 (human flow map generation unit), and a sensor arrangement adjustment unit 5 (sensor arrangement determination unit). The sensor installation position support device 1 is a device that automatically outputs a sensor arrangement with the best measurement sensitivity in a target area based on each information.

  First, the outline of each function shown in FIG. 1 will be described. The measurement model generation unit 3 determines the measurement model indicating the measurement range of the sensor from the sensor information 102 and the measurement item information 103. The human flow map generation unit 4 determines the measurement area. The function of creating a human flow map indicating the flow of people in the target area based on the map information 100 and the human flow information 101. The sensor arrangement adjustment unit 5 checks the measurement model against the human flow map and determines the sensor position at which the measurement sensitivity of the target area is the best. It has a function to output. Note that the target area is a range that the customer wants to measure with a sensor. For example, in the case of a camera, it indicates the area to be monitored, and in the case of a laser radar, the area to be measured. Hereinafter, details of each function will be described.

<Measurement model generator>
FIG. 2 is a diagram illustrating an example of sensor characteristic information acquired by the measurement model generation unit 3. FIG. 3 is a diagram illustrating an example of a measurement model generated by the measurement model generation unit 3. The measurement model generation unit 3 will be described with reference to FIGS.

  FIG. 2 is an example of sensor characteristic information acquired by the measurement model generation unit 3 from the sensor information 102. The measurement accuracy of each sensor, the measurement accuracy according to the distance from the sensor to the measurement person, and the density of persons It shows the characteristics of the measurement accuracy (resolution) according to the degree and the items (measurement items) that can be measured by the sensor. Note that the sensor and the characteristic information are not particularly limited thereto, and may be characteristic information in which the range, accuracy, and resolution of the same sensor vary according to the measurement item. Further, the measurement accuracy does not necessarily have to be different depending on the distance, and may be characteristic information in which the accuracy of the entire measurement range is uniform.

  The measurement model generation unit 3 generates a sensor measurement model as shown in FIG. 3 by collating customer measurement needs, such as person tracking and suspicious person detection, extracted from the measurement item information 103 with sensor characteristic information. In FIG. 3, reference numeral 10 denotes a monocular camera, 11a and 11b denote examples of a measurement model at the time of human tracking and object detection by the monocular camera, 12 denotes a laser sensor, and 13a and 13b denote measurement models at the time of human tracking and object detection by the laser sensor. An example is shown.

  It should be noted that the shape of the measurement model does not need to be changed depending on the measurement items as shown in FIG. Factors that use the method of changing the shape of the model depending on the measurement items include, for example, human tracking, which recognizes objects as people and follows them, compared to object detection, which detects objects in the target area. Is difficult, and there is a difference in the measurement range and accuracy. Therefore, even in the same sensor, the size of the shape of 11a and 13a is smaller than 11b and 13b. In general, a camera uses a measurement model whose accuracy changes according to the distance from the sensor 10, such as 11a or 11b, in order to express characteristics such as difficulty in detecting an object that is too close or a distant object. You may. The shape of the measurement model is not particularly limited to the shape shown in FIG. Further, the information used to generate the measurement model is not limited to the information on the characteristics of the sensors and the measurement items, but may use map information in the building, for example, the height of the installation location and the depression angle. May be used to make the shape of the measurement model variable.

  That is, the measurement model generation unit 3 (measurement model generation means) uses at least one of the layout information, the flow direction information, the position environment information, and the congestion state information of the target area to measure the measurement range or / and the measurement of the measurement model. The accuracy may be changed. Here, the layout information of the target area includes map information, the height of the building, the entrance of the building, and the like. The flow direction information includes a flow map generated by the flow map generation means. The position environment information includes the environment of the measurement position such as backlight or indoor / outdoor. The congestion status information is information indicating whether or not the location is easily crowded, such as congestion or idleness.

<People flow map generator>
FIG. 4 is a diagram illustrating the processing of the human flow map generation unit 4 (see FIG. 1). FIG. 5 is a flowchart showing the processing of the human flow map generation unit 4. The flow map generator 4 will be described with reference to FIGS. 4 and 5.

  First, the human flow map generation unit 4 obtains the layout overhead view 20 (layout information) of the target area shown in FIG. 4 from the map information 100 (process S41). It is divided into small areas (process S42). Next, the human flow data 22 in the target area is extracted from the human flow information 101 (process S43), and the direction of the main human flow in each small area is estimated (process S44). Finally, a flow map 23 reflecting the direction of the flow of people in the overhead view of the layout is output (process S45). Hereinafter, this small area is called a cell and will be described.

  As a method of dividing the cells in the layout overhead view 10 in the processing S42, the size of the cells may be smaller than the measurement model, and it is not necessary that all cells have the same size. The shape is not particularly limited as long as it can cover the entire bird's-eye view in addition to the lattice shape such as 21.

  The human flow data acquired in the process S43 may be any information that can estimate the main direction of the human flow in the target area, such as the trajectory information of the person. For example, if there is already a person who actually moves in the target area, data obtained by temporarily arranging sensors for measuring the flow of people is obtained.If no person exists, the doorway or corner in the layout bird's-eye view The data may be human flow data estimated by simulation from the above information or the like, and is not particularly limited.

  In the process S44, the direction 24 of the human flow in each grid of the layout overhead view is determined from the obtained human flow data 22. As a method of estimating the direction 24 of the human flow, there is a method of collating the human flow data 22 with 21 and estimating a direction vector having the most frequent occurrence with the center of the grid as the origin. In the present embodiment, the final direction of the human flow is determined by setting the direction of the human flow to four patterns as in 24 and comparing the direction vector estimated in each grid to which pattern is similar. . Note that the number of patterns in the direction of the human flow is not limited to four, and the accuracy of sensor arrangement can be improved as the number of patterns is increased.

<Sensor arrangement adjustment unit>
FIG. 6 is a diagram showing functional blocks of the sensor arrangement adjusting unit 5. The sensor arrangement adjusting unit 5 will be described with reference to FIG. The sensor arrangement adjustment section 5 has functions of a measurement center point detection section 30, a measurement model collation section 31, and a sensor arrangement position output section 32. To explain the outline of each function, the measurement center point detection unit 30 determines the important measurement point indicating the measurement center position in the target area from the information of the human flow map, and the measurement model collation unit 31 determines the acquired measurement model as important. The function of updating the measurement sensitivity of all cells in the target area by collating the measurement point with the human flow map. The sensor arrangement position output unit 32 outputs the sensor position information from the optimized sensitivity map information. It has a function to do. Hereinafter, details of each function will be described.

  FIG. 7 is a diagram illustrating an example of important measurement points (candidate measurement points) output by the measurement center point detection unit 30. As a method of determining an important measurement point, a search window 40 including four cells is scanned, and two conditions of “all cells include direction information of a human flow” and “variance of the direction of a human flow are maximum” are set. There is a method of searching for four adjacent cells to be filled and setting the center position as an important measurement point (candidate measurement point).

  The reason for the former condition is to implement an efficient sensor arrangement by excluding areas such as walls, which are one of the layout information, where there is no person, from the measurement target.The latter reason is to grasp human flow information This is because when a person or a person is to be tracked, a point (congestion status information) where the direction of the flow of people (flow direction information) changes a lot is likely to be an important measurement target.

  In addition, as a method of determining an important measurement point, a method of determining the position of an entrance of a building, which is positional environment information from map information, or a method of manually setting by a user using a GUI (Graphical User Interface) or the like The method is not particularly limited as long as one or more important measurement points can be determined. The measurement method of the important measurement point may be switched according to the measurement item. For example, when the measurement item is person tracking, the method shown in FIG. 7 may be used, and when the suspicious individual is detected, the method of determining the position of the entrance of a building which is considered to be the central area of measurement from the map information may be used.

  That is, the measurement center point detection unit (important measurement point detection unit) performs measurement in the target area using at least one of the layout information, the flow direction information, the position environment information, and the congestion state information of the area to be monitored. One or more important points to be extracted may be extracted.

  FIG. 8 is a flowchart illustrating the processing of the measurement model matching unit 31. First, the measurement model matching unit 31 acquires a measurement model from the measurement model generation unit 3 (process S1). Subsequently, the measurement sensitivity of the cell in the vicinity of the important measurement point detected by 30 is confirmed (processing S2), and if less than the threshold (processing S2, No), the measurement model is collated with the important measurement point (S3), and is equal to or more than the threshold If it is (process S2, Yes), after creating a new important measurement point around the cell where the measurement sensitivity is not set, the process S3 is performed (process S4). Then, the measurement sensitivity value of each cell whose measurement model has been collated is updated (process S5), and it is confirmed whether the measurement sensitivity of all cells in the target area is equal to or larger than the threshold (process S6). If it is less than the threshold (process S6, No), the process returns to process S2, and if it is more than the threshold (process S6, Yes), the process of the measurement model collating unit 31 ends. Then, the sensor arrangement position output 32 (see FIG. 6) outputs the sensor arrangement position. Hereinafter, the method of matching the measurement model in step S3, the method of creating important measurement points in step S4, and the method of calculating measurement sensitivity in step S5 will be described in detail.

  FIGS. 9A and 9B are diagrams for explaining the process S3 of the measurement model collation unit 31. FIG. 9A is a diagram illustrating a human flow map, FIG. 9B is a diagram illustrating the measurement direction of the measurement model, and FIG. The process S3 will be described with reference to FIG. Here, a description will be given of a flow of collating a camera measurement model with a person flow map 51 in which important measurement points 50 shown in FIG. 9A are set when the measurement item is person tracking.

  9B is an enlarged view of the vicinity of the important measurement point 50, 53 is a camera, 54 is a measurement model of 53, 55 is a center line of the measurement model, and 56a, 56b, and 56c are people flows around 50. , 57 indicates an overlapping portion of the area outside the target area and the area of the measurement model, and 58 indicates a straight line connecting the sensor installation position and the important measurement point.

  In the process S3, first, an installation position of the measurement model 54 in the human flow map is determined, and then an optimal measurement direction is estimated based on the direction of the human flow. As a method of determining the installation position of 54, there is a method of adjusting the center of gravity of the measurement model to the important measurement point 50. In addition, the method of setting the center of gravity of the area with the highest measurement accuracy in the measurement model, or adding the position information with the highest measurement accuracy obtained in advance from sensor information etc. to the measurement model when generating the measurement model However, there is a method of matching the position with an important measurement point, and the method is not particularly limited.

  Next, a method for estimating the measurement direction of the measurement model will be described. When a person is tracked by a camera, a person moving in a vertical direction rather than a horizontal movement with respect to the camera tends to be easier to track. In addition, there is an advantage that the face of the suspicious individual is easily photographed by photographing the suspicious individual in the vertical direction.

  Therefore, in this example, a direction in which angles θa, θb, and θc formed by the center line 55 of the measurement model and the straight lines 56a, 56b, and 56c indicating the directions of the respective people are less than or equal to a certain value is set as a measurement model direction candidate. The direction in which the overlapping area 57 between the non-measurement area such as a wall and the measurement model is the minimum is defined as the direction of the final measurement model.

  When the installation position of the camera 53 is set to an area that is not an installation target, such as a wall or outside a person flow map, from the collated measurement model, the installation position and the important measurement point are represented by a straight line 58 (FIG. 9C). In conclusion, the measurement model may be translated so that the installation position is included in the installation target. In the case where the optimum measurement model position is output as a plurality of candidates by the above-described method, a method of automatically selecting one of them or a method of manually selecting one of them is used. In order to determine the measurement direction of the measurement model, map information or the like may be referred to.For example, priority is given to the direction in which the doorway can be photographed in order to emphasize photographing of people entering and exiting, and measurement due to disturbance such as backlighting. In order to prevent a decrease in accuracy or protect privacy, a method of giving priority to a direction in which the outdoors are not photographed may be adopted.

  FIGS. 10A and 10B are diagrams for explaining the processing S5 of the measurement model collating unit 31. FIG. 10A shows the measurement model 60, FIG. 10B shows the measurement models 60a and 60b collated with the human flow map 51, and FIGS. It is an enlarged view of the cell in the people flow map 51. Before describing the process S4, a process S5 for calculating the measurement sensitivity of the cell will be described with reference to FIG.

  In FIG. 10A, reference numeral 60 denotes a measurement model, and in FIG. 10B, reference numerals 60a and 60b denote measurement models collated with the human flow map 51 in step S3, and reference numerals 61a and 61b denote enlarged views of two cells in the human flow map. ing. In FIG. 10 (c), in the cell 61a, only the measurable area of the measurement model 60b exists, 62 indicates an area with a measurement accuracy of 90%, and 63 indicates an area with a measurement accuracy of 70%. Also, in FIG. 10D, in the cell 61b, there are both measurable regions of the measurement models 60a and 60b, 64 is a region of 60% of the measurement accuracy of 60a, and 65 is a region of the measurement accuracy of 40% of 60b. An area 66 indicates an overlap area where the measurement accuracy of 60a and 60b is 40%. The measurement sensitivity when there is no overlap with other regions such as the regions 62, 63, 64, and 65 can be derived from [Equation 1].

  When there is an overlap as in the region 66, it can be derived by calculating the average of the measurement sensitivities obtained respectively in [Equation 1]. Then, the sum of the measurement sensitivities of the respective regions is the measurement sensitivity of the cell including the region. The method of calculating the measurement sensitivity is not particularly limited as long as the measurement sensitivity can be calculated for each cell, and the value is determined in consideration of the measurement accuracy of the sensor and can be relatively compared between the cells.

  FIG. 11 is a diagram illustrating the process S4 of the measurement model matching unit 31. The processing S4 will be described with reference to FIG. In FIG. 11, reference numeral 70 denotes an example of a human flow map on which the measurement sensitivity is reflected, reference numeral 71 denotes an important measurement point, reference numeral 72 denotes a measurement sensitivity value of each cell, reference numeral 73 denotes a threshold value of the measurement sensitivity existing near the important measurement point (30 in this example). ), Cells 74a, 74b, 74c, 74d, and 74e indicate candidate positions for newly created important measurement points. In the process S4, the positions of the important measurement points are determined so that many of the 73 measurement sensitivities are equal to or more than the threshold value under the condition set in advance.

  As a method of determining the position, as shown in the example of FIG. 11, a grid point where the measurement sensitivity has already been calculated and the contact point 73 are selected as candidates, “all four surrounding cells are within the target area”, “important measurement points” There is a method of outputting the final position on the basis of a condition such as "the distance is short". In this example, 74c is a position candidate.

  Note that the position candidates are not limited to grid points, and may be center points of cells, grid lines, or the like, or may use map information such as "prioritize cells close to the entrance", and are particularly limited to the above conditions. Not done.

  In the present embodiment, by repeatedly performing the processing S4 and the processing S3, the measurement sensitivity of each cell is updated, and the sensor position where the measurement accuracy of the entire target area is optimized is output. Therefore, when performing the process S3, the measurable region of the measurement model already matched with the important measurement point may overlap with the measurable region of the measurement model adjusted to the new important measurement point. The position and orientation of the measurement model may be adjusted.

  FIG. 12 is a diagram illustrating an example in which the measurement model is compared by the measurement model matching unit 31. For example, as shown in FIG. 12, when the measurement model 80b is collated with the important measurement point 81 obtained in the process S4, an overlapping area 82 with the measurement model 80a occurs. In this case, for detection of a suspicious person or the like, it is required to install a sensor so that a blind spot does not occur in the target area. Therefore, it is preferable to increase the priority of the measurement method in which the overlapping area 82 is reduced. In addition, in the case of tracking a person, since the measurement accuracy increases as the number of overlapping regions 82 increases, the priority in the measurement direction in which the overlapping region 82 has an area equal to or greater than a certain ratio of the measurement model may be increased. In addition, a method of adding the ratio of the overlapping area 82 in each cell to a condition for determining the measurement method may be used, and there is no particular limitation.

  The sensor arrangement position output unit 32 (see FIG. 6) outputs the final sensor arrangement position from the position information of the measurement model in the human flow map acquired by the measurement model collation unit 31.

  FIG. 13 is a diagram illustrating an example of a sensor position output by the sensor arrangement adjustment unit 5. Through the processing described above, the sensor arrangement adjustment unit 5 determines the positions of all sensors at each important measurement point in the target area as shown in FIG.

  Although the sensor arrangement adjusting unit 5 (sensor arrangement determining means) of the present embodiment determines the sensor arrangement position based on the measurement model and the human flow map, the present invention is not limited to this. The sensor arrangement adjusting unit 5 may determine the arrangement position of the sensor using at least one of the layout information, the flow direction information, the position environment information, and the congestion state information of the target area.

  As the measurement model used in the sensor arrangement adjustment unit 5, a model having the same shape but a different shape may be used. For example, since the height at which the sensor is arranged can be obtained from the position of the measurement model that has been collated with the map information and the important measurement point, the measurement model generated from the height information may be collated with the flow map in step S3.

  In addition, a more accurate sensor is determined by judging the target area where congestion is likely to occur or is quiet from the human flow information and comparing the measurement model in which measurement accuracy has changed and calculating the measurement sensitivity of each cell within that area. Can be determined.

  Note that the sensor arrangement adjustment unit 5 may use measurement models of different sensors. For example, basically, the measurement model of the camera is checked against the flow map, and if the flow direction of the cell is the same in a wide area, it is determined that the area is unlikely to cause congestion, and the measurement model of the laser radar is used. A method of collation may be used. In this case, for example, the number of cameras and laser radars to be used or the total cost may be designated by the user in advance, and the conditions may be added as optimization conditions.

  When the sensor arrangement adjustment unit 5 uses measurement models of different sensors, for example, in the case of the process S3 in FIG. 8, when the overlap rate of the measurement range of the collated measurement model and the measurement range of the new model is large, There is a method of adding a process of changing the measurement model to select the one with the smaller overlapping rate, and minimizing the number of installed sensors.

  Further, the sensor arrangement pattern outputted by the sensor arrangement adjusting unit 5 is not limited to one pattern, and a plurality of patterns may be outputted. For example, when determining the sensor position at a certain important measurement point, if the same measurement sensitivity is shown in two measurement directions, select one and output the sensor arrangement, then select the other A method may be used in which the sensor arrangement is output and the user finally selects the sensor arrangement, or a method of automatically selecting the higher one by comparing the calculated measurement sensitivities of the entire target area.

  In the first embodiment of the present invention, a measurement model corresponding to the characteristics of the sensor and the measurement items is generated by the above-described functional configuration, and the measurement model is compared with a flow map created from layout information of the measurement range to calculate the measurement sensitivity. In addition, the arrangement positions of a plurality of sensors can be automatically output.

  In the first embodiment, the camera installation position can be output using information not only in the two-dimensional space but also in the three-dimensional space. For example, instead of a human flow map in which a two-dimensional map is divided into cells, a human flow map in which a three-dimensional space is divided into a plurality of voxels is used, and the direction and measurement sensitivity of the human flow in each voxel are derived using a three-dimensional measurement model. Thus, the series of flows in FIG. 1 may be performed.

<< Embodiment 2 >>
FIG. 14 is a functional block diagram of the sensor installation position support device 1A according to the second embodiment. The sensor installation position support device 1A has a sensor information acquisition unit 6 and a person flow information acquisition unit 7 as compared with FIG. 1, and the database 2 stores the sensor installation information 104 acquired by the sensor information acquisition unit 6. ing. In the second embodiment, in a situation where a sensor has already been installed, correction of a sensor installation method and output of a new sensor arrangement position are performed.

  The sensor arrangement adjustment unit 5A calculates the entire measurement sensitivity from the information of the existing sensors in the target area, determines an area with low measurement sensitivity, and corrects the sensor installation method to improve the sensitivity, In some cases, a new sensor installation position is output. Hereinafter, details of the sensor arrangement adjustment unit 5A will be described.

  FIG. 15 is a flowchart illustrating a process of the sensor arrangement adjustment unit according to the second embodiment. In FIG. 15, processes S2, S4, S5, and S6 are the same as the processes in FIG. First, the sensor arrangement adjustment unit 5A calculates the measurement sensitivities of all the cells in the target area from the human flow map of the target area and the measurement model of the currently installed sensor (step S20). Next, one or more initial positions of important measurement points in the target area are determined by using the same method as that of the measurement center point detection unit 30 in FIG. 6 (process S21).

  Then, the sensor arrangement adjustment unit 5A determines whether or not the measurement sensitivity of the cell near the important measurement point is equal to or more than the threshold (process S2). If the measurement sensitivity of the cell near the important measurement point is equal to or larger than the threshold (Yes in step S2), a new important measurement point is created (step S4), and the process proceeds to step S22. If the measurement sensitivity of the cell near the important measurement point is less than the threshold (No in step S2), the process proceeds to step S22.

  In the process S22, it is determined whether or not an existing measurement model (a measurement model of a currently installed sensor) exists near an important measurement point. Specifically, the distance between the important measurement point and the position of the existing measurement model is calculated, and if it is within the threshold value, it is determined that it is close. It should be noted that the confirmation method of the process S22 is not particularly limited as long as the method can compare the positional relationship.

  If it is determined in step S22 that there is no existing measurement model (step S22, No), a new measurement model is acquired from the measurement model generation unit 3 and the important measurement model is processed in the same manner as in step S3 in FIG. (Step S23), and the process proceeds to Step S5.

  On the other hand, when it is confirmed in step S22 that an existing measurement model exists (step S22, Yes), the existing measurement model is collated with an important measurement point (step S24), and the process proceeds to step S5. In step S24, only the measurement direction is adjusted because the position of the measurement model is fixed, and the method may be the same as step S3.

  Lastly, the measurement sensitivity of the cell whose measurement model has been collated is updated (step S5). In step S6, if the measurement sensitivity of all cells in the target area is equal to or greater than the threshold (step S6, Yes), the processing is terminated. When the processing is completed and is less than the threshold value (No in step S6), the process returns to step S2.

  In the second embodiment, the position of the measurement model in the human flow map at the time when the processing is terminated by the sensor arrangement adjusting unit 5A is output as the final sensor arrangement position. In addition, in the sensor arrangement adjusting unit 5A, depending on the user's request, a process of creating a new measurement model and collating is not performed, and only the measurement model of the existing sensor is collated with the human flow map so that the measurement direction of the sensor is changed. A correction method may be used.

  FIGS. 16A and 16B are diagrams illustrating an example of a sensor position adjusted by the sensor arrangement adjusting unit 5A, wherein FIG. 16A illustrates the measurement sensitivity before adjustment, and FIG. 16B illustrates the measurement sensitivity after adjustment. FIG. 16 illustrates an example in which the sensor installation angle is adjusted by the sensor arrangement adjustment unit 5A. In FIG. 16, 53 is a sensor, 60c is a measurement model of 53, 70a and 70b are human flow maps reflecting measurement sensitivities before and after sensor position adjustment by the sensor arrangement adjustment unit 5A, and 110 is one cell in the target area. Is shown. For example, if the user sets the threshold value of the measurement sensitivity to 50, the measurement sensitivity of the cell 110 is insufficient, so that the measurement sensitivity of the cell 110 is changed to 50 by changing the installation angle of the cell 110, and all cells in the target area are changed. It is possible to output a sensor arrangement in which the measurement sensitivity of the cell is equal to or more than a threshold.

  According to the second embodiment of the present invention, by the above-described functional configuration, the measurement sensitivity is calculated by comparing the measurement model with the human flow map in consideration of the position information of the sensor already installed in the target area, and the existing sensor is The installation method can be modified, and the installation position of a new sensor can be output if necessary.

  In the second embodiment, when creating a person flow map, creating a measurement model, and collating the measurement model with important measurement points, detection of the person flow information and the person measured by the existing sensor acquired by the person flow information acquisition unit 7 is performed. A result such as a rate may be used. For example, when the flow of people changes significantly due to a layout change in the facility, compared to before the change, by applying the second embodiment, a sensor arrangement suitable for the layout after the change can be output.

  The sensor installation position support device 1 according to the present embodiment includes a measurement model generation unit 3 that generates a measurement model indicating a measurement range of the sensor based on characteristic information and measurement item information of the sensor, map information of an area to be monitored, and A human flow map generation unit 4 that generates a human flow map indicating the direction of human flow in the area based on the human flow information of the area, and a measurement sensitivity for each cell constituting the area based on the measurement model and the human flow map, and the calculated value is calculated. A sensor arrangement adjusting unit for determining a sensor arrangement position in the area based on the measurement sensitivity. The measurement model includes the measurement accuracy according to the distance from the sensor and the measurable range. When the ratio of the measurement accuracy to 1 is A, the measurement accuracy is A for the cell area. Is a value obtained by integrating the ratio of the surface line of A with A.

  By applying the sensor installation position support devices 1 and 1A of the above-described embodiment, the installation positions of one or a plurality of sensors can be automatically determined according to the measurement needs of the customer.

1,1A Center installation position support device 2 Database (storage means)
3 Measurement model generation unit (measurement model generation means)
4 Traffic flow map generator (People flow map generator)
5,5A sensor arrangement adjustment unit (sensor installation position determining means)
6 Sensor information acquisition unit 7 People flow information acquisition unit 10 Monocular camera (sensor)
12 Laser sensor (sensor)
11a, 11b, 13a, 13b Measurement model 20 Layout overhead view (layout information)
23 People flow map 100 Map information 101 People flow information 102 Sensor information 103 Measurement item information 104 Sensor installation information

Claims (11)

A measurement model generation unit that generates a measurement model indicating a measurement range of the sensor based on the characteristic information and the measurement item information of the sensor;
Means for generating a people flow map indicating the direction of the people in the area based on the map information of the area to be monitored and the people information of the area,
The measurement model and on the basis of the human stream map to calculate the measurement sensitivity for each cell constituting the area, have a sensor arrangement determining means for determining the position of the sensor in the area based on the measurement sensitivity the calculated,
The sensor arrangement determination unit compares the measurement accuracy included in the measurement model with the flow map, calculates the measurement sensitivity for each cell,
The measurement model includes a measurement accuracy and a measurable range according to a distance from the sensor,
Wherein the measurement sensitivity is a value obtained by integrating the ratio of the area of A to the area of the cell and the value of A when the ratio of the measurement accuracy to 1 is A. Support equipment. The people flow map generation means obtains layout information of the area from the map information of the area, divides the obtained layout information into cells constituting the area, and calculates a flow direction of each cell from the flow information. The sensor installation position support device according to claim 1, wherein the determination is made. The sensor arrangement determining means extracts at least one candidate measurement point to be measured in the area using at least one of the area layout information, the flow direction information, the location environment information, and the congestion state information, The sensor installation position support device according to claim 1, wherein the measurement model is arranged at the extracted candidate measurement point. The said sensor arrangement | positioning determination means determines the arrangement | positioning position of the said sensor using at least one or more of the layout information of the area, the flow direction information of a person, the location environment information, and the congestion situation information. A sensor installation position support device as described in the above. The measurement model generating means changes a measurement range or / and a measurement accuracy of the measurement model using at least one of layout information, flow direction information, position environment information, and congestion state information of the area. The sensor installation position support device according to claim 1. The sensor arrangement determining means, when the measurement sensitivity of the cells in the vicinity of the candidate measurement point is greater than a predetermined threshold value, according to claim 3, characterized in that the vicinity of the candidate measurement point and the candidate measurement point of the new 3. The sensor installation position support device according to 1. 4. The sensor installation position support device according to claim 3 , wherein the sensor arrangement determination unit changes the measurement model such that measurement sensitivity of a cell in the area is equal to or more than a predetermined threshold. 5. 2. The sensor installation position support device according to claim 1, wherein when there are a plurality of arrangement positions of the sensors, the sensor arrangement determination unit uses measurement models of different types of sensors. 3. The sensor installation position assistance device further includes a sensor information acquisition unit that acquires installed sensor installation information,
The sensor installation position support device according to claim 3 , wherein the sensor arrangement determination unit sets the sensor installation position acquired by the sensor information acquisition unit as the candidate measurement point. The sensor according to claim 9 , wherein the measurement model generation unit generates a measurement model indicating a measurement range of the sensor based on the characteristic information and the measurement item information of the sensor acquired by the sensor information acquisition unit. Installation position support device. The sensor installation position support device further has a human flow information acquisition unit that obtains human flow information from the installed sensors,
The sensor installation position support device according to claim 1, wherein the information on the area is set as the information on the person flow based on the information on the person flow acquired by the information acquisition unit on the person flow.

Images