WO2024014001A1 - Detection device, detection method, and detection program - Google Patents

Abstract

A detection device 10 is equipped with: an acquisition unit 101 for acquiring coordinate-equipped event occurrence data in which an event occurrence point is recorded and polygon map data which includes a polygon which makes it possible to identify a location on a map; and an output unit 103 for determining whether an event has occurred in a polygon and outputting the determination results, on the basis of whether or not the event occurrence point is located inside a polygon of the polygon map data.

Description

Detection device, detection method, and detection program

The disclosed technology relates to a detection device, a detection method, and a detection program.

As disclosed in Non-Patent Document 1, there is research that calculates hot spots where police events frequently occur.

It is also effective to shorten the travel time from receiving notification that an event has occurred to the point where the event occurred. However, if the event is a police event, it is preferable to focus on the area where the event is occurring in advance and take action, such as patrolling, in the area of interest to prevent the event from occurring itself. Similarly, even if the event is not a police event, by paying attention to the area where the event is occurring in advance, you can shorten the travel time from receiving notification that the event has occurred to the location where the event occurred. becomes possible.

In Non-Patent Document 1 mentioned above, kernel density estimation is used to calculate hot spots, which are locations where events frequently occur, based on points where events occur. In kernel density estimation, event hotspots are indicated by a range based on a pre-given bandwidth (fixed value), so when different locations are concentrated in a narrow area such as a built-up area, it is difficult to determine which locations ( It is difficult to determine where the outbreak occurred (in a building). Furthermore, in cases where one type of place is spread over a wide area, such as a park, it is necessary to identify events that occur at distant points even within the same park as having occurred at the same place (park). is difficult.

The disclosed technology has been made in view of the above points, and provides a detection device and a detection method that detect event hotspots in a narrower range while detecting events that occur at the same location as the same hotspot. , and a detection program.

A first aspect of the present disclosure is a detection device, and an acquisition unit that acquires event occurrence data with coordinates in which an event occurrence point is recorded, and polygon map data including polygons that can identify a location on a map. and an output unit that determines whether the event occurred in the polygon based on whether or not the point of occurrence of the event is inside the polygon in the polygon map data, and outputs the determined result.

A second aspect of the present disclosure is a detection method, in which a processor acquires event occurrence data with coordinates in which an event occurrence point is recorded, and polygon map data including polygons that can identify a location on a map. Then, depending on whether the occurrence point of the event is inside the polygon in the polygon map data, it is determined whether the event occurred in the polygon or not, and a process of outputting the result is executed.

A third aspect of the present disclosure is a detection program that causes a computer to function as the detection device of the first aspect.

According to the disclosed technology, it is possible to provide a detection device, a detection method, and a detection program that detect event hotspots in a narrower range and detect events that occur at the same location as the same hotspot. .

FIG. 2 is a block diagram showing the hardware configuration of a detection device according to an embodiment of the disclosed technology. FIG. 2 is a block diagram showing an example of a functional configuration of a detection device. FIG. 3 is a diagram showing an example of event occurrence data. It is a figure showing an example of road network data. FIG. 2 is a diagram showing a map that visualizes road network data. FIG. 3 is a diagram for explaining a method of calculating the distance between a point with event ID: 1001 and a road section with section ID: 2. It is a diagram showing the result of visualizing the shortest distance d between section ID: 3 of road ID: 1234 and event ID: 1001. It is a diagram showing the result of visualizing the relationship between section ID: 1 of road ID: 5678 and event ID: 1001. It is a diagram showing a state in which the road ID, section ID, and distance with the shortest distance for each event are added to the event occurrence data. It is a figure which shows the result of the output part counting the number of events for each section ID of each road ID. FIG. 6 is a diagram illustrating an example in which the output unit highlights a section ID of a corresponding road ID based on the number of counted events. It is a flowchart which shows the flow of detection processing by a detection device. FIG. 3 is a diagram showing an example of event occurrence data. FIG. 3 is a diagram showing an example of a map for recording event occurrence data. 14 is a diagram showing the result of plotting the position of the event occurrence data shown in FIG. 13. FIG. FIG. 3 is a diagram showing an example of polygon map data stored as polygon data. FIG. 3 is a diagram showing an example of polygon types used for hot spot detection. FIG. 3 is a diagram illustrating an example of assigning hot spot IDs to polygon map data. FIG. 3 is a diagram showing an example of visualization of hot spot detection results. FIG. 3 is a diagram illustrating an example of a precise area of a hot spot. 20 is a diagram showing the results of hot spot calculation visualized in FIG. 19. FIG. It is a flowchart which shows the flow of detection processing by a detection device. FIG. 3 is a diagram illustrating an example in which a hot spot is ambiguously detected. FIG. 3 is a diagram illustrating an example of a hot spot detection area based on kernel density estimation. FIG. 6 is a diagram illustrating an example of a case where no hot spot is detected. FIG. 6 is a diagram illustrating an example of a case where no hot spot is detected. It is a flowchart which shows the flow of detection processing by a detection device. FIG. 3 is a diagram illustrating an example of event occurrence data to which a polygon type is assigned. 29 is a diagram showing the results of calculating the proportion of polygon types (locations) where an event occurs using the event occurrence data of FIG. 28. FIG. It is a figure showing an example of facility information. FIG. 7 is a diagram illustrating an example of the results of determining which polygon the facility information falls into. FIG. 3 is a diagram illustrating an example in which facility category information is added to event occurrence data. FIG. 7 is a diagram illustrating an example of calculating occurrence rate information for each facility category for each event category. 34 is a diagram illustrating an example of visualizing the processing result of FIG. 33. FIG. 34 is a diagram illustrating an example of visualizing the processing result of FIG. 33. FIG. 34 is a diagram illustrating an example of visualizing the processing result of FIG. 33. FIG.

Hereinafter, an example of an embodiment of the disclosed technology will be described with reference to the drawings. In addition, the same reference numerals are given to the same or equivalent components and parts in each drawing. Furthermore, the dimensional ratios in the drawings are exaggerated for convenience of explanation and may differ from the actual ratios.

Although each of the embodiments below will be explained using an example of patrol, the present invention is not limited to patrol, but can also be used for taxis, ride sharing, delivery, etc.

(First example)
FIG. 1 is a block diagram showing the hardware configuration of a detection device 10 according to a first example of this embodiment. The detection device 10 of the first embodiment acquires event occurrence data in which the point of occurrence of an event is recorded and road network data in which information on a road consisting of one or more sections is recorded, and each of the acquired data. This is a device that detects and outputs locations where events frequently occur based on the content. The event occurrence data may further include information on the date and time of occurrence of the event.

Note that, for example, a general-purpose computer device such as a server computer or a personal computer (PC) can be applied to the detection device 10 according to the present embodiment.

As shown in FIG. 1, the detection device 10 includes a CPU (Central Processing Unit) 11, a ROM (Read Only Memory) 12, a RAM (Random Access Memory) 13, a storage 14, an input section 15, a display section 16, and a communication interface. interface ( I/F) 17. Each configuration is communicably connected to each other via a bus 19.

The CPU 11 is a central processing unit that executes various programs and controls various parts. That is, the CPU 11 reads a program from the ROM 12 or the storage 14 and executes the program using the RAM 13 as a work area. The CPU 11 controls each of the above components and performs various arithmetic operations according to programs stored in the ROM 12 or the storage 14. In this embodiment, the ROM 12 or the storage 14 stores a detection program for detecting and outputting points where events frequently occur.

The ROM 12 stores various programs and various data. The RAM 13 temporarily stores programs or data as a work area. The storage 14 is constituted by a storage device such as an HDD (Hard Disk Drive) or an SSD (Solid State Drive), and stores various programs including an operating system and various data.

The input unit 15 includes a pointing device such as a mouse and a keyboard, and is used to perform various inputs.

The display unit 16 is, for example, a liquid crystal display, and displays various information. The display section 16 may adopt a touch panel method and function as the input section 15.

The communication interface 17 is an interface for communicating with other devices. For this communication, for example, a wired communication standard such as Ethernet (registered trademark) or FDDI, or a wireless communication standard such as 4G, 5G, or Wi-Fi (registered trademark) is used.

Next, the functional configuration of the detection device 10 will be explained.

FIG. 2 is a block diagram showing an example of the functional configuration of the detection device 10.

As shown in FIG. 2, the detection device 10 has an acquisition section 101, a distance calculation section 102, and an output section 103 as functional configurations. Each functional configuration is realized by the CPU 11 reading a detection program stored in the ROM 12 or the storage 14, loading it into the RAM 13, and executing it.

The acquisition unit 101 acquires event occurrence data in which the point of occurrence of an event is recorded, and road network data in which information about a road consisting of one or more sections is recorded. Note that the event occurrence data may further record information on the date and time of occurrence of the event.

FIG. 3 is a diagram showing an example of event occurrence data. In this embodiment, the event occurrence data includes information such as an event ID, the latitude and longitude at which the event occurred, the date and time of receipt, the day of the week, the event category, and whether the event occurred indoors or outdoors. It is assumed that event occurrence data will be listened to and recorded over the telephone when a report is received, such as at a police command division. In particular, it is assumed that location information will be entered by clicking on a rough location on an electronic map with a mouse cursor when listening. Of course, a record of the location information obtained directly from the informant who visited the police box or police station may be kept.

The premise is that on days of the week, times of day, and locations where an event has occurred multiple times in the past, there is a tendency for the probability that the same event will occur in the future to be higher.

FIG. 4 is a diagram showing an example of road network data. In this embodiment, the road network data includes a road ID (information that identifies a road, corresponding to National Highway No. X, Prefectural Road No. , longitude 1 (longitude of the start point of section ID), latitude 1 (latitude of the start point of section ID), longitude 2 (longitude of the end point of section ID), latitude 2 (latitude of the end point of section ID), width (in meters).

The distance calculation unit 102 calculates, for each road, the distance between the occurrence point of each event in the event occurrence data and each section obtained from the road network data. A distance calculation method by the distance calculation unit 102 will be explained.

FIG. 5 is a diagram showing a map that visualizes the road network data shown in FIG. 4. Using a map that visualizes road network data, we will explain a method for calculating the distance between the occurrence point of each event in the event occurrence data and each section obtained from the road network data.

The distance calculation unit 102 first draws a line segment from each past event occurrence data point to the start point and end point of the road section, and a line segment is created by these line segments and the line segment of the road section. Calculate the angle. FIG. 6 is a diagram for explaining a method of calculating the distance between the point with event ID: 1001 and the section of the road with section ID: 2 of road ID: 1234. Angle A is an angle formed by a line segment drawn from the point of event occurrence data of event ID: 1001 to the start point of section ID: 2 of road ID: 1234, and a line segment of section ID: 2. Angle B is an angle formed by a line segment drawn from the point of event occurrence data of event ID: 1001 to the end point of section ID: 2 of road ID: 1234, and a line segment of section ID: 2. Since the coordinates of the three points forming each of angles A and B are known, these angles can be calculated from the inner product formula.

If the angles of angle A and angle B are both less than 90 degrees, the distance calculation unit 102 calculates the shortest distance d between the coordinates of event ID: 1001 and section ID: 2. Since various methods have already been proposed for finding the shortest distance between a line segment and a point, the distance calculation unit 102 may select any one of the methods to calculate the distance d.

FIG. 7 is a diagram showing the results of visualizing the shortest distance d between section ID: 3 of road ID: 1234 and event ID: 1001. As a result of comparing the shortest distance d between section ID: 2 and event ID: 1001 with the shortest distance d between section ID: 3 and event ID: 1001, the shortest distance to section ID: 2 is shorter, so the section ID:2 remains as a candidate for the road section closest to the point of event ID:1001.

FIG. 8 is a diagram showing the result of visualizing the relationship between section ID: 1 of road ID: 5678 and event ID: 1001. As shown in FIG. 8, in this case, one angle is 90 degrees or more, so the distance calculation unit 102 calculates the shortest distance between section ID: 1 of road ID: 5678 and event ID: 1001. is not calculated.

In this way, the distance calculation unit 102 calculates the shortest distance d between all road sections that satisfy the angle condition and the coordinates of the event, and calculates the section ID with the shortest distance d from among them, and the road ID. is specified, and information on the specified road ID and section ID is added to past event occurrence data.

FIG. 9 is a diagram showing a state in which the road ID, section ID, and distance with the shortest distance for each event are added to the event occurrence data as a result of distance calculation by the distance calculation unit 102.

Note that the distance calculation unit 102 may limit the road network data for which the distance is to be calculated to road network data having coordinates within a certain distance from the point of the event occurrence data to be calculated. Further, the distance calculation unit 102 divides the road network data in advance into a lattice shape using a quadratic mesh of a map, etc., and selects the nearest road section by limiting it to the road network data of the area that includes the point of the event occurrence data to be calculated. You can ask for it. These limitations allow the detection device 10 to reduce distance calculation time.

The output unit 103 outputs the number of events for which the distance calculated by the distance calculation unit 102 is the shortest for each section. A method of outputting the number of events by the output unit 103 will be explained.

The output unit 103 counts the number of events associated in the event occurrence data of FIG. 9 for each section ID of the road ID. FIG. 10 is a diagram showing the result of counting the number of events for each section ID of each road ID by the output unit 103. The output unit 103 may output the table shown in FIG. The output unit 103 outputs the number of events for which the distance calculated by the distance calculation unit 102 is the shortest for each section, thereby identifying a road section that should be given priority and attention because many events occur nearby. It can be output.

Furthermore, the output unit 103 may highlight the section ID of the corresponding road ID based on the counted number of events. FIG. 11 is a diagram showing an example in which the output unit 103 highlights the section ID of the corresponding road ID based on the counted number of events. In FIG. 11, past event data is also displayed together with roads so that the meaning of the processing can be understood, but in reality, individual occurrence points may not be displayed. This is because the amount of actual event occurrence data is enormous, and displaying individual event occurrence points would result in poor visibility.

The distance calculation unit 102 may use all the data recorded in the event occurrence data for the event to calculate the shortest distance to the road segment, or may use the data after filtering by the time of occurrence, event category, day of the week, etc. It may be processed with For example, if the indoor/outdoor column is given the condition "indoor", the deterrent effect may be low even if the patrol is carried out, so the distance calculation unit 102 determines that the indoor/outdoor column is given the "indoor" condition. The distance may be calculated by excluding events that occur.

Furthermore, depending on the type of event category, it is thought that there are cases where a high deterrent effect can be expected through patrols and cases where it cannot be expected, so the distance calculation unit 102 filters according to the event category and then calculates the section of road that has the shortest distance. You can ask for it. In the example of the event occurrence data shown in FIG. 3, if it is considered that mere patrolling is not expected to have much of a deterrent effect on shoplifting, the distance calculation unit 102 filters out shoplifting and only applies to events other than shoplifting. You can perform the process using

Even if the shortest road section is identified for each event, if the distance is longer than a certain level, we believe that patrolling that section will have a low deterrent effect. You can also find the number. For example, taking the distance column in FIG. 9 as an example, if it is considered that patrolling the shortest road will be less effective if the distance is 5 m or more, the output unit 103 will output a distance of 5 m or more. All you need to do is to exclude those that are present and then find the section of the road that should be emphasized. Furthermore, if the width of the road seems to be an influence, the output unit 103 may add the width of the road to the distance calculated by the distance calculation unit 102, and then determine whether the distance exceeds a predetermined threshold. good.

Note that although road network data is used in this embodiment, the present disclosure is not limited to such an example. The detection device 10 may use information on boundary lines between roads and sidewalks, roads and buildings, etc. in the map data.

The output unit 103 further lists and ranks the distances between the event occurrence point and each road section in descending order of distance, assigns points according to the ranking, and highlights the road sections based on the score calculation results. You can. For example, the output unit 103 sets points as 3 points, 2 points, and 1 point for the first to third places in order of distance, calculates the score of each road section, and assigns a section based on the calculation result. may be highlighted. In this case, the output unit 103 may perform display such as changing the thickness of the section or changing the color of the section depending on the score, for example.

The output unit 103 may solve a so-called traveling salesman problem to find the shortest route that passes through all of the highlighted road sections, and present the route. By patrolling according to the route output by the output unit 103, effective patrolling becomes possible. However, since the traveling salesman problem is an NP-hard problem, as the scale increases, it is difficult to obtain an exact solution, and a local solution must be found using a greedy method or a local search method.

Although the explanation has been given using patrol as an example, the present disclosure is not limited to such an example. The present disclosure can also be used for calls for services such as taxis, deliveries, etc. to call vehicles by telephone. The present disclosure can also be used to meet the demand for services that do not involve calling a vehicle, such as obtaining location information by means other than the telephone, such as delivery and presentation of information according to location. As the location information, for example, location information issued by a mobile information terminal used by the user may be used. For example, if you use data on points where users have called in the past and use the method described above to find out which roads are closest to the points where those calls were made, you can improve the efficiency of service operation. We can hope that this can be achieved. For example, the results of the techniques described above may be used in optimization problems such as the traveling salesman problem described above.

The detection device 10 may combine a method of detecting hot spots where events occur frequently using kernel density estimation and a method of detecting hot spots as road sections. For example, by detecting only events that are considered to occur on the road as road section hotspots, the detection device 10 makes it easier to understand what should be noted when patrolling road section hotspots, and makes it easier to carry out the patrol. It becomes easier.

Next, the operation of the detection device 10 will be explained.

FIG. 12 is a flowchart showing the flow of detection processing by the detection device 10. The detection process is performed by the CPU 11 reading the detection program from the ROM 12 or the storage 14, expanding it to the RAM 13, and executing it.

In step S101, the CPU 11 acquires event occurrence data in which the point of occurrence of the event is recorded, and road network data in which information about a road consisting of one or more sections is recorded.

Following step S101, in step S102, the CPU 11 calculates the distance between the occurrence point of each event in the event occurrence data and each section in the road network data for each road. The distance calculation method in step S102 is the same as the distance calculation method by the distance calculation unit 102 described above.

Following step S102, the CPU 11 outputs the number of events for which the calculated distance is the shortest for each section in step S103. The method of outputting the number of events for which the calculated distance is the shortest in step S103 is the same as the method of outputting by the output unit 103 described above.

As explained above, according to the detection device 10 of the present embodiment, it is possible to detect a road section that should be given priority attention because many events occur nearby.

(Second example)
The detection device 10 of the second embodiment acquires event occurrence data with coordinates in which the event occurrence point is recorded, and polygon map data that can identify locations such as buildings, parks, roads, etc., and This device determines whether an event occurrence point is included in a polygon of a location that can be used for spot detection, and outputs the determination result.

FIG. 13 is a diagram showing an example of event occurrence data. The event occurrence data includes an event ID, the latitude and longitude at which the event occurred, the date and time of receipt, the day of the week, the event category, and the like. It is assumed that this event occurrence data will be listened to and recorded over the telephone when a report is received, such as at a police command division. Of course, event occurrence data may also be recorded by interviewing a caller who visits a police box or police station in person. In particular, position information may be input by the person recording the event occurrence data plotting the rough position on the electronic map with a mouse cursor when hearing from the informant. FIG. 14 is a diagram showing an example of a map for recording event occurrence data.

FIG. 15 is a diagram showing the results of plotting the positions of the event occurrence data shown in FIG. 13. Although they are omitted from the electronic maps in Figures 14 and 15 because they reduce visibility, actual electronic maps include building addresses or names, and the person recording the event occurrence data is informed by the informer. It is possible to identify a location that matches the address or building name confirmed over the phone.

Polygon map data is data in which areas such as roads, various buildings, parks, planting areas, parking lots, etc. are stored as polygon data. FIG. 16 is a diagram showing an example of polygon map data stored as polygon data. Each polygon in the polygon map data is assigned a polygon ID and a polygon type. Each vertex forming a polygon has latitude and longitude coordinates, and if the latitude and longitude are connected in the order of the vertex ID of the same polygon ID, and finally the first vertex ID is connected, the polygon is formed.

It is also assumed that on days of the week, times of day, and places where a specific event has occurred multiple times in the past, there is a tendency for the probability that the same event will occur in the future to be higher.

The acquisition unit 101 acquires event occurrence data with coordinates and polygon map data that can identify locations such as buildings, parks, roads, etc. The event occurrence data with coordinates is data as shown in FIG. 13, for example. The output unit 103 extracts event occurrence data with coordinates, for example, for each event category, and determines which polygon of the polygon map data each coordinate of the event occurrence data falls inside. Further, the event occurrence data with coordinates may further record information on the date and time of occurrence of the event. If the coordinate-attached event occurrence data records information on the event occurrence date and time, the output unit 103 extracts the coordinate-attached event occurrence data for a certain period of time (for example, for the past year), for example, for each event category. , it may be determined which polygon of the polygon map data each coordinate of the event occurrence data falls inside. Note that in the second embodiment, the distance calculation unit 102 shown in the first embodiment is not an essential component.

Various methods have already been devised to determine whether or not the object is inside a complex-shaped polygon, so the output unit 103 only needs to use one of the methods. One of the simplest methods is to rasterize polygon data to create bitmap data, fill the inside with a specific color, and change the coordinates of past events to the coordinates on the bitmap data at the same ratio as the polygon data. This method determines whether the coordinates are filled with a specific color when converted.

For example, if polygon map data is layered, such as a park or planting area polygon existing under a certain building polygon, the coordinates of individual events are determined to be inside multiple polygons. It should be done. In such a case, the output unit 103 may determine the priority order of the layers and determine that the layer is located inside the layer with the higher priority order. For example, if the priority order is determined in the order of building and planting area, if the output unit 103 determines that the area is inside the polygon of both the building and the planting area, the output unit 103 determines that the area is inside the polygon of the building. You may judge.

There are times when it is better to detect hot spots where events occur frequently based on map polygons, and times when it is not. FIG. 17 is a diagram showing an example of polygon types used for hot spot detection. In the example of FIG. 17, buildings and parks are polygon types used for hot spot detection.

Then, the output unit 103 clusters only the event occurrence data determined to be inside the building and park polygons for each same polygon. Event occurrence data belonging to each cluster is assigned a hotspot ID, for example, with P added to the beginning of the corresponding polygon ID. FIG. 18 is a diagram illustrating an example of assigning hot spot IDs to polygon map data. In the example of FIG. 18, event IDs 1001, 1002, and 1007 are inside polygon ID: 111. Therefore, the output unit 103 assigns a hotspot ID of P111 to events having event IDs of 1001, 1002, and 1007. In this embodiment, the letter P is added to the beginning of the hotspot ID, but it goes without saying that the letter or symbol added to the beginning of the hotspot ID is not limited to P. Letters or symbols may be added.

The event occurrence data shown in Figure 17, which is determined to be inside a polygon of a polygon type different from the polygon type used for hot spot detection, indicates that the shape of the polygon is not considered to be effective for hot spot detection. be. Therefore, the output unit 103 may detect hot spots using kernel density estimation or the like, which is commonly used in this field. Specifically, the output unit 103 determines the bandwidth of kernel density estimation (the range affected by each sample). Then, the output unit 103 may create a Gaussian distribution or the like up to the band width centering on the position of each sample and superimpose them, and may designate a region where the value exceeds a certain value as a hot spot. If a building or park is a polygon type used for hot spot detection, event occurrence data that was inside a road polygon, etc. is applicable. The output unit 103 assigns a unique ID consisting only of numbers and no characters or symbols at the beginning to each event of the event occurrence data that was inside a road polygon or the like.

Through the processing up to this point, we have been able to detect multiple hot spots where the probability of an event occurring is considered to be high. FIG. 19 is a diagram showing an example of visualization of hot spot detection results. In the example shown in FIG. 19, a shopping mall, a park, some buildings in a bar district, and two locations on a road are detected as hot spots. However, the two locations on the road with hot spot IDs 10000 and 10001 are hot spots detected by the output unit 103 through kernel density estimation, and the circles around the hot spots represent the bandwidth of kernel density estimation. Then, the exact area of the hot spot is the area inside the boundary of the overlapping circles where the integration of the Gaussian distribution has a value greater than a certain value. FIG. 20 is a diagram showing an example of a precise area of a hot spot. In the example of FIG. 20, the areas indicated by dotted lines with hotspot IDs 10000 and 10001 are the exact hotspot areas.

FIG. 21 is a diagram showing the results of the hot spot calculation visualized in FIG. 19, in which the hot spot ID is added to the event occurrence data shown in FIG. 13.

Next, the operation of the detection device 10 will be explained.

FIG. 22 is a flowchart showing the flow of detection processing by the detection device 10. The detection process is performed by the CPU 11 reading the detection program from the ROM 12 or the storage 14, expanding it to the RAM 13, and executing it.

In step S111, the CPU 11 obtains event occurrence data in which the point of occurrence of the event is recorded, and polygon map data that can identify locations such as buildings, parks, roads, etc.

Following step S111, in step S112, the CPU 11 determines which polygon in the polygon map data each event in the event occurrence data has occurred within. Specifically, the CPU 11 determines whether the coordinates of each event in the event occurrence data are included inside a polygon in the polygon map data.

Following step S112, the CPU 11 outputs the determination result in step S112 in step S113. When outputting the determination result, the CPU 11 may output it in a visualized state by superimposing it on the polygon map data as shown in FIG. It may be output with columns added.

The effects of this embodiment will be explained once again. Conventionally, in kernel density estimation, which is most commonly used in hot spot detection, the bandwidth (the range affected by each sample) must be given as a fixed value. The range of the hotspot changes depending on the value of this bandwidth. If this value is large, for example, there is a high possibility that a hot spot in a large park or a large building can be appropriately detected from past event occurrence data. On the other hand, in places where small buildings are densely packed, such as in a bar district, there are only a few shops (buildings) that often receive complaints or crime reports (in fact, they often show such a tendency). , hotspots are vaguely detected across multiple buildings, making it impossible to narrow down the buildings that require attention.

FIG. 23 is a diagram showing an example in which a hot spot is ambiguously detected. In FIG. 23, a circle is drawn around the position of each event. This circle represents the bandwidth based on a certain fixed value. In addition, in FIG. 23, the bandwidth for events on the road is omitted because visibility becomes poor. In kernel density estimation, after overlaying Gaussian distributions etc. that follow this shape, a region where the density exceeds a certain threshold is defined as a hot spot. In other words, the area inside the overlapping circles becomes a hot spot. FIG. 24 is a diagram illustrating an example of a hot spot detection area based on kernel density estimation. If the bandwidth is increased in this way, in the area on the lower left where small buildings are clustered, buildings where no event has occurred will also be detected as hot spots.

On the other hand, if the bandwidth is reduced, in an area where small buildings are densely packed, only specific buildings where events occur frequently can be detected as hot spots. However, if a large area building or park, such as a large shopping mall or large park, is a true hotspot, then the circles for individual events representing bandwidth are likely not to overlap. Therefore, there is a very high possibility that hot spots will not be detected.

FIGS. 25 and 26 are diagrams showing an example of a case where no hot spot is detected. By reducing the bandwidth, the circles for individual events representing the bandwidth do not overlap in the shopping mall in the upper left and the park in the upper right, so these locations cannot be detected as hot spots.

This embodiment solves the problem of hot spot detection using kernel density estimation. The detection device 10 can determine the future probability of occurrence of each detected hot spot using information on the occurrence time of past event data. Here, it is sufficient to use an appropriate prediction model according to the characteristics of the target event. The simplest predictive model is that the probability of occurrence at this hotspot follows a Poisson distribution. When using the Poisson distribution, the expected value can be found by dividing the past number of occurrences by the observation period. For example, in the example shown in FIG. 21, the noise complaint occurrence data with the hot spot ID of P112 is 3 times in 110 days from July 1, 2021 to October 18, 2021. Therefore, when calculating the expected value that will occur on October 19, 2021, it will be 3/110=0.027.

On the other hand, it has been reported that a model called close repeated victimization is effective for many types of crimes. This is a model that assumes that when a crime occurs, there is a high possibility that the same crime will occur again immediately after or in the vicinity. Depending on the type of crime, such a model may be applied to determine the future probability of occurrence. Here, it is also possible to consider further processing for quantitatively evaluating the plurality of hot spots detected above.

Chainey et al. proposed PAI (Prediction Accuracy Index) as an evaluation index for hotspots (Spencer Chainey, Lisa Tompson and Sebastian Uhlig, The Utility of Hotspot Mapping for Predicting Spatial Patterns of Crime, Security Journal, 2008, 21, (4 - 28)). This can be expressed by the following formula.
PAI = (Number of events at the place where the event is expected to occur/Total number of events in the target area)/(Area of the place where the event is expected to occur/Area of the total target area) ... (1)

In PAI, the locations where an event is expected to occur are detected in a more limited manner, and the score becomes higher if the prediction is correct. If the process of this embodiment is applied here, the area of the entire target area of the above formula (1) is the area of the entire map in FIG. 19, and the area of the place where the event is expected to occur is the hotspot ID: P111, The total area of P112, P201, P205, P210, 10000, 10001, the total number of events in the target area, the number of events that actually occurred on the entire map in Figure 19 during the evaluation period, and the events in the locations where the events are expected to occur. The number may be calculated as the number of events that occurred within the hotspot out of the number of events that actually occurred.

In the above example, kernel density estimation was used to detect hotspots of event data that was determined to be inside a polygon of a polygon type other than the polygon type used for hotspot detection, as shown in Figure 17. Disclosure is not limited to such examples. Below, an example will be described in which hot spot detection of event data determined to be inside a polygon of a polygon type other than the polygon type used for hot spot detection is calculated at a higher speed.

FIG. 27 is a flowchart showing the flow of detection processing by the detection device 10. The detection process is performed by the CPU 11 reading the detection program from the ROM 12 or the storage 14, expanding it to the RAM 13, and executing it.

In the detection process, the detection device 10 first prepares an array (cluster array) for storing a plurality of clusters. One cluster is a set of a plurality of past event data, and the cluster arrangement is an arrangement of the plurality of past event data sets. Initially, the number of elements in the cluster array is 0.

In step S201, the CPU 11 sorts only the event data determined to be inside a polygon of a polygon type that is not the polygon type used for hot spot detection in descending order of receipt date and time t.

The CPU 11 scans the data starting from the oldest reception date and time t, and in step S202 calculates which cluster in the array of clusters the coordinates of each event data are closest to. Information on the center of gravity of each cluster may be used for this determination. The center of gravity of each cluster is the average of the coordinates of past event data included in that cluster.

Following step S202, the CPU 11 determines whether the distance to the nearest cluster is less than a threshold value in step S203.

As a result of the determination in step S203, if it is less than the threshold (step S203; Yes), the CPU 11 adds the event data to the cluster in step S204.

On the other hand, if the result of the determination in step S203 is that it is not less than the threshold (step S203; No), the CPU 11 creates a new cluster containing only the event data in step S205, and adds it to the cluster array.

Following step S204 or step S205, the CPU 11 determines whether scanning of all event data has been completed in step S206.

As a result of the determination in step S206, if scanning of all event data is not completed (step S206; No), the CPU 11 advances the reception date and time t to the next reception date and time in step S207, and proceeds to the process of step S202. return.

On the other hand, as a result of the determination in step S206, if scanning of all event data has been completed (step S206; Yes), a cluster consisting of a set of multiple events or a cluster consisting of only one event is determined in the cluster arrangement. will be stored in. Here, in step S208, the CPU 11 finally excludes the cluster consisting of one event, and the remaining individual clusters are events that are determined to be inside a polygon of a polygon type other than the polygon type used for hot spot detection. Make it a data hotspot. The area of the hot spot obtained in this way may be a rectangle using the minimum latitude and longitude and the maximum latitude and longitude of the coordinates of event data belonging to the cluster. Note that if there is data with exactly the same reception date and time, the CPU 11 may repeat the processing from step S202 to step S205 for the number of data pieces.

In this embodiment, the detection device 10 detects hot spots based on polygons for buildings and parks, and detects hot spots for roads using another method, but the present disclosure does not apply to such an example. Not limited.

For example, the detection device 10 may detect hot spots on the road based on polygons. As an example of such a case, a road polygon having an area of a certain size or more is divided in advance into units such as intersections, and the detection device 10 detects hot spots on the road using the method of this embodiment. If we can successfully divide polygons, we can expect to be able to detect hotspots in a narrower range than when we use kernel density estimation to detect hotspots on roads.

For example, as another embodiment, a polygon having an area of a certain amount or more is divided into a plurality of polygons in advance, and the detection device 10 detects hot spots on the road by the method of this embodiment. good. For example, when detecting a certain large park as a hot spot, the park is divided into a plurality of polygons in advance, and the detection device 10 uses the method shown in this embodiment on the divided polygons. Also good. Thereby, the detection device 10 can be expected to be able to detect smaller hot spots instead of one large hot spot.

As another example, the detection device 10 does not detect a polygon as one hot spot, but detects a hot spot using kernel density estimation or the like for only event data within one polygon. You can. Thereby, the detection device 10 can be expected to be able to detect a more appropriate hot spot when the occurrence of an event is concentrated in a part of the polygon.

As another example, the detection device 10 may detect hot spots using kernel density estimation or the like on event data inside the polygon only for polygons having an area of a certain size or more. For example, the detection device 10 can be used to detect polygons whose area is less than a certain threshold value for the entire polygon, and for polygons whose area exceeds the threshold value, it can be used to narrow down the area to be detected. We can expect that this will become possible.

As another example, a large commercial facility may contain multiple stores, and a park may consist of a baseball field, playground equipment, sandbox, etc. . If it is possible to use information on polygons inside the polygon, such as the shape of each store, the detection device 10 may divide the entire polygon into smaller shapes or use only the information on the inside polygons. Therefore, hot spots may be detected by applying the method of this embodiment or kernel density estimation. By dividing the entire polygon into smaller shapes or using only the information on the inner polygons, the detection device 10 is expected to be able to detect even finer hot spots than when using the entire polygon as is. .

The detection device 10 may detect hot spots by kernel density estimation with a variable bandwidth depending on the size of the polygon to which the occurrence point belongs. Thereby, the detection device 10 can prevent a situation where a hot spot is ambiguously detected as shown in FIG. 23 or a situation where a hot spot is not detected as shown in FIGS. 25 and 26. An example of how to determine the bandwidth is to set a relatively large bandwidth for occurrence points that are within a polygon with a large area that exceeds a predetermined threshold, and set a relatively large bandwidth for occurrence points that are within a polygon that has a small area that is less than or equal to a predetermined threshold. A possible method is to set a smaller bandwidth for the occurrence point than the bandwidth set for the occurrence point within a polygon with a large area. Further, the detection device 10 may determine the bandwidth so as to be proportional to the area of the polygon, or may determine the bandwidth by nonlinearly converting the area of the polygon using a sigmoid function or the like.

(Third example)
Similar to the second embodiment, the detection device 10 of the third embodiment acquires event occurrence data with coordinates in which the point of occurrence of the event is recorded, and polygon map data that can identify locations such as buildings, parks, roads, etc. This device determines whether an event occurrence point is included in a polygon of a location that can be used for hot spot detection in polygon map data, and outputs the determination result. In the third embodiment, the detection device 10 calculates the event occurrence rate in the location or facility category for each event category. The detection device 10 calculates the event occurrence rate in a location or facility category for each event category, thereby making patrol activities more efficient.

The acquisition unit 101 acquires event occurrence data with coordinates and polygon map data that can identify locations such as buildings, parks, roads, etc. The event occurrence data with coordinates is data as shown in FIG. 13, for example. The output unit 103 calculates the event occurrence rate in the location or facility category for each event category based on the data acquired by the acquisition unit 101, and outputs the calculation result.

An example of the processing of the output unit 103 will be explained. As in the second embodiment, the output unit 103 extracts event occurrence data with coordinates, for example, for each event category, and determines which polygon of the polygon map data each coordinate of the event occurrence data falls inside. . Further, the event occurrence data with coordinates may further record information on the date and time of occurrence of the event. If the coordinate-attached event occurrence data records information on the event occurrence date and time, the output unit 103 extracts the coordinate-attached event occurrence data for a certain period of time (for example, for the past year), for example, for each event category. , it may be determined which polygon of the polygon map data each coordinate of the event occurrence data falls inside. The output unit 103 then adds the polygon type of the determined polygon to the event occurrence data. FIG. 28 is a diagram illustrating an example of event occurrence data to which a polygon type is assigned.

Next, the output unit 103 uses the event occurrence data assigned the polygon type to calculate the proportion of polygon types (locations) where the event occurs. FIG. 29 is a diagram showing the results of calculating the proportion of polygon types (locations) where events occur using the event occurrence data of FIG. 28. In the example of FIG. 29, it can be seen that only "fights" occur in two polygon types (locations): buildings and roads. From this, it can be seen that when performing patrol activities for a certain event category, patrolling can be carried out efficiently by focusing on the type or proportion of the revealed occurrence locations.

It is also important for efficient patrols to confirm the facility category in which events that occur within buildings occur. The output unit 103 uses the facility category information and the facility information with coordinates to extract in what facility category the event that occurs in the building occurs. FIG. 30 is a diagram showing an example of facility information. The facility information is data created in advance by the user.

The output unit 103 uses the facility information to determine which polygon each facility is located inside. FIG. 31 is a diagram showing an example of the result of determining which polygon the facility information falls into.

The output unit 103 adds facility category information to the event occurrence data contained in the same polygon. FIG. 32 is a diagram showing an example in which facility category information is added to event occurrence data.

Then, the output unit 103 calculates information on the occurrence rate for each facility category for each event category, and adds it to the calculation results shown in FIG. 29 for visualization. FIG. 33 is a diagram showing an example of calculating occurrence rate information for each facility category for each event category.

By calculating the output unit 103 in this way, when carrying out patrol activities for a certain event category, it is possible to calculate not only the type and proportion of the occurrence location that has been clarified previously, but also the occurrence location for each facility category for buildings. By referring to the ratio information, patrols can be carried out more efficiently. For example, in the example shown in FIG. 33, the facility category with the highest incidence of shoplifting is convenience stores, so if convenience stores are concentrated on vigilance, the possibility of preventing the occurrence of incidents increases.

34 to 36 are diagrams showing examples of visualizing the processing results of FIG. 33. The output unit 103 outputs visualized processing results as shown in FIGS. 34 to 36, for example. Those conducting patrol activities should check the visualization results in advance before conducting patrols. FIG. 36 is an example in which the number of incidents of shoplifting by facility category is visualized as is, instead of the column for the facility category ratio of shoplifting in FIG. 33. In this way, the output unit 103 may visualize the number of occurrences by facility category as is, without calculating the percentage.

(Fourth example)
The detection device 10 may combine the processing shown in the first embodiment and the processing shown in the second embodiment. That is, the detection device 10 may combine the road section hot spot detection method of the first embodiment with the polygon-based hot spot detection method of the second embodiment. For example, the detection device 10 may detect hot spots for buildings and parks using the process shown in the second embodiment, and may detect hot spots for roads using the process shown in the first embodiment. As a result, the detection device 10 can understand that the hot spot in the road section requires attention on the road, which may make it easier to carry out patrols.

(Fifth example)
The detection device 10 may combine the processing shown in the second embodiment and the processing shown in the third embodiment. That is, the detection device 10 applies the method of calculating the event occurrence rate for each location or facility category of the third embodiment in addition to the polygon-based hot spot detection method of the second embodiment. You can. The detection device 10 can output locations to be patrolled intensively by detecting hot spots and calculating the event occurrence rate.

Although the embodiments of the present disclosure have been described above in detail with reference to the accompanying drawings, the technical scope of the present disclosure is not limited to such examples. It is clear that a person with ordinary knowledge in the technical field of the present disclosure can come up with various changes or modifications within the scope of the technical idea described in the claims. It is understood that changes or modifications of the above naturally fall within the technical scope of the present disclosure.

Furthermore, the effects described in the above embodiments are explanatory or exemplary, and are not limited to those described in the above embodiments. In other words, the technology according to the present disclosure can be obtained from the descriptions in the above embodiments together with the effects described in the above embodiments, or in place of the effects described in the above embodiments, and has common knowledge in the technical field of the present disclosure. It may have other effects that are obvious to those who are interested in it.

Note that the detection processing executed by the CPU reading the software (program) in each of the above embodiments may be executed by various processors other than the CPU. The processor in this case is a PLD (Programmable Logic Device) whose circuit configuration can be changed after manufacturing, such as an FPGA (Field-Programmable Gate Array), and an ASIC (Application Specific Intel). In order to execute specific processing such as egrated circuit) An example is a dedicated electric circuit that is a processor having a specially designed circuit configuration. Further, the detection process may be executed by one of these various processors, or by a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, a CPU and an FPGA, etc.). ) can also be executed. Further, the hardware structure of these various processors is, more specifically, an electric circuit that is a combination of circuit elements such as semiconductor elements.

Furthermore, in each of the above embodiments, a mode has been described in which the detection program is stored (installed) in the storage 14 in advance, but the present invention is not limited to this. The program can be installed on CD-ROM (Compact Disk Read Only Memory), DVD-ROM (Digital Versatile Disk Read Only Memory), and USB (Universal Serial Bus) stored in a non-transitory storage medium such as memory It may be provided in the form of Further, the program may be downloaded from an external device via a network.

Regarding the above embodiments, the following additional notes are further disclosed.

(Additional note 1)
memory and
at least one processor connected to the memory;
including;
The processor includes:
Obtaining event occurrence data with coordinates in which the point of occurrence of the event is recorded, and polygon map data including polygons that can identify the location on the map,
A detection device configured to determine and output whether or not an event occurs in a polygon based on whether or not a point where the event occurs is inside the polygon in the polygon map data.

(Additional note 2)
A non-transitory storage medium storing a program executable by a computer to perform a detection process,
The detection process includes:
Obtaining event occurrence data with coordinates in which the point of occurrence of the event is recorded, and polygon map data including polygons that can identify the location on the map,
A non-temporary storage medium that determines and outputs whether or not an event has occurred in a polygon based on whether or not the occurrence point of the event is inside the polygon in the polygon map data.

10 Detection device 101 Acquisition unit 102 Distance calculation unit 103 Output unit

Claims (8)

1. an acquisition unit that acquires event occurrence data with coordinates in which the event occurrence point is recorded, and polygon map data including polygons that can identify locations on the map;
   an output unit that determines and outputs whether or not the event occurred in the polygon based on whether or not the occurrence point of the event is inside the polygon in the polygon map data;
   A detection device comprising:
2. The polygon map data further includes a polygon type that identifies a location category of the polygon;
   2. The detection device according to claim 1, wherein the output unit determines a point where an event frequently occurs, limited to a specific polygon type.
3. The event occurrence data further records the date and time of occurrence of the event,
   For the polygon types other than a specific polygon type, the output unit is arranged such that each event is an event at an occurrence point that is less than a predetermined threshold from the center of gravity of a past occurrence point, in the order of the occurrence date and time. 3. The detection device according to claim 2, which determines and outputs whether or not the event has occurred in the polygon type.
4. The acquisition unit further acquires road network data in which information about a road consisting of one or more sections is recorded,
   For the polygon types other than a specific polygon type, the output unit calculates the distance between the point of occurrence of the event and each of the sections for each road, and calculates the distance calculated for each of the sections. Outputs the number of events for which is the shortest,
   The detection device according to claim 2.
5. The output unit divides the polygon into a predetermined number of parts, and determines whether the event occurs in the divided polygon based on whether the occurrence point of the event is inside the divided polygon. 2. The detection device according to claim 1, wherein the detection device outputs the following information.
6. The detection device according to claim 2, wherein the output unit outputs a proportion of the polygon types in which the event occurs in the polygon map data.
7. The processor
   Obtaining event occurrence data with coordinates in which the point of occurrence of the event is recorded, and polygon map data including polygons that can identify the location on the map,
   A detection method that executes a process of determining and outputting whether or not an event occurs in a polygon based on whether or not the occurrence point of the event is inside the polygon in the polygon map data.
8. A detection program for causing a computer to function as the detection device according to any one of claims 1 to 6.