KR100818334B1 - Detecting method at automatic police enforcement system of illegal-stopping and parking vehicle and closed-circuit television for crime prevention and system thereof - Google Patents

Abstract

A method and a system for automatically detecting illegal-stopping and parking vehicles using a closed-circuit television for preventing crimes are provided to enhance the performance of the system by minimizing calculation amounts of morphology and labeling. Calculation amounts of image analysis are minimized by using an ROI(Region Of Interest) among whole regions obtained by a camera(30). After removing shades and modeling background images, images are formed as blocks in a vehicle detecting module so as to minimize calculation amounts of morphology and labeling. Morphology calculation is sequentially executed to remove a hole in divided regions. The labeling is then executed for region division. An MER(Minimum Enclosed Rectangle) calculation is then executed. Moving objects are detected by estimating the MER calculation. By receiving the moving object information, police regulation on illegal-stopping and parking vehicles is executed through a P/T camera(32).

Description

Detecting Method at Automatic Police Enforcement System of Illegal-stopping and Parking Vehicle and Closed-Circuit Television for Crime Prevention and System

The present invention relates to an unmanned automatic control method and system having an illegal parking and crime prevention closed circuit television function, and more particularly, to a remote location including a security closed circuit television (CCTV) function based on the illegal parking unmanned automatic regulation. Unattended automatic crackdown with illegal parking and crime prevention closed circuit television function by detecting the vehicle by video analysis that detects the movement of the vehicle in the parking lot monitoring area and conducts automatic crackdown. It relates to a method and a system thereof.

Illegal parking stops the normal traffic flow of the vehicle on the road with heavy traffic, narrows the road width, and can be a major cause of personal and object traffic accidents. Therefore, it is necessary to crack down on illegal parking vehicles in areas that prohibit parking.

As such, the crackdown on vehicles illegally parked on the roadside can be classified into two types. First, the first type is an artificial crackdown by illegal parking enforcement personnel, and the second type is a remote crackdown using a surveillance camera. In addition, remote control using a surveillance camera can be divided into two types again. In other words, the subject of the control is a manual method by human operation, and the subject of the control is an automatic method by the operation of the system.

In the case of the artificial crackdown method, the crackdown is performed directly by the crackdown personnel. However, at present, the area for parking is absolutely insufficient compared to the number of domestic vehicles, and the current situation is that the enforcement personnel cannot be placed in every crackdown area for crackdowns. As a result, even though the parking is prohibited, illegal parking is frequently caused by the driver's disregard of the law. After the crackdown service was over, it was troublesome to return to the crackdown agency and re-enter the data about the caught vehicle.

In order to solve the problem of the artificial crackdown method, a surveillance camera is installed in a crackdown site around some roads, and a remote crackdown is adopted using the surveillance camera thus installed. However, most of the cases in which the subject who controls the video information of a plurality of remotely installed surveillance cameras in the central situation room are artificially manual, and the other cases of the automatic method expose the following problems.

First, the vehicle detection method by image analysis of the image transmitted from the surveillance camera involves a large amount of calculation process. By setting the target area of the image analysis to the whole area of the parking monitoring area, unnecessary calculation process is performed as much as the difference between the parking monitoring area and the prohibited parking area.

Second, in the case of a system using the simple image transfer method and the difference image between the background and the current image frame as the vehicle detection method among the image analysis methods, the vehicle detection rate and the accuracy of detection are inferior because it cannot adapt to the environmental change in the crackdown site. It is true. In addition, when the movement of the vehicle moves very slowly, when the difference image technique between the previous and current image frames is used, a fatal error or a defect occurs because there is no movement of the object. In addition, a motion vector is generated due to the noise of the surveillance camera, which is often incorrectly recognized as a moving object.

Third, in the case of the simple moving object detection method excluding the shadow elimination that causes the inaccurate feature in the image analysis, only the existence of the moving object is judged, so it is determined whether the moving object is a vehicle or an object other than the vehicle, that is, a person or an animal. Could not. Therefore, the detection of incorrect moving object involves unnecessary operation of the system.

Fourth, in the conventional method using the vehicle detection method by image analysis, a large amount of computation is required in morphology and labeling operations that require image processing in units of pixels, thereby improving the accuracy and stability of the overall system operation. Lowers. Therefore, there is a need for a method of minimizing the amount of computation in both operations.

Fifth, since there was no way to automatically crack down and handle the vehicles parked before the parking enforcement began, this problem could be exploited, causing many problems.

The present invention is to solve the problems of the prior art, firstly, the present invention does not set the target region of the image analysis to the entire area of the parking monitoring area, but limited to a plurality of Regions of Interest (ROI) unnecessary area It is an object of the present invention to provide an unmanned automatic control method and a system having illegal parking and crime prevention closed-circuit television functions for eliminating unnecessary system operation due to the calculation of and the result of image analysis of unnecessary area.

In addition, the present invention divides the image into n × n block units by image block method and performs morphology and labeling operations by block unit instead of pixel unit, thereby minimizing the calculation amount of two operations and real-time vehicle detection and vehicle. Another object of the present invention is to provide an unmanned automatic control method and system having illegal parking and crime prevention closed circuit television functions capable of improving the detection and tracking success rate.

In order to achieve the above object, the present invention receives an image of an intermittent or surveillance site photographed in real time through a fixed camera for detection installed on a road, into an ROI setting module, and sets an ROI module for each main region of interest of the image. Illegal parking and crime prevention closed-circuit television function by image analysis to set multiple regions of interest in ROI, remove shadow area from shadow removal module, and update background model in adaptive background modeling module In the unmanned automatic crackdown method having (A), a region of interest (ROI) is performed without performing image analysis on the entire area of the image of the crackdown site, which is photographed in real time through a fixed camera for detection in the entire process of image analysis. To minimize the amount of computational analysis by limiting only the region of interest to the target region for image analysis Setting a region of interest for the user; (b) blocking the image in the vehicle detection module to minimize the amount of morphology and labeling calculation in the image analysis process after removing the shadow area and modeling the background image: (c) a hole in the partition after the image blocking sequentially performing morphology operations of expansion and erosion to remove holes; (d) performing labeling for region division after the morphology operation; (e) performing a MER (minimum neighboring rectangle) calculation after the labeling operation; (f) detecting a moving object by evaluating each MER representing each divided area after the MER calculation; (g) controlling an unattended parking control and crime prevention by controlling the control P / T camera by receiving the moving object detection information, and performing an unmanned automatic control method having illegal closed parking and crime prevention closed circuit television functions. It is characterized by the provision.

In addition, the present invention, the fixed camera for detecting a parking stop area is installed in the parking lot monitoring area is divided into the monitoring sector on both sides of the road in proportion to the distance; A region of interest setting module for setting a plurality of regions of interest (ROIs), which are vehicle detection regions for image analysis, of an image of an enforcement scene photographed in real time through the fixed camera for detection; An initial background model is produced from 1 to N images sequentially input from the fixed camera for detection, and the difference between the produced initial background model and the N + 1th input image is the moving image and the shadow of the moving object. A shadow removal module dividing a foreground area and removing a shadow area of a moving object from the foreground area; In the shadow removing module, the moving object area from which the shadow area is removed is taken as the detected moving object area, and the background including the previously tracked moving object area is excluded except for the moving object area detected in the N + 1th input image. An adaptive background modeling module for updating the model; The shadow removal module illegally divides the image into blocks of a certain size and performs an morphology and labeling operation in blocks larger than the pixel unit to perform image blocking to minimize the amount of illegal parking and crime prevention. It is characterized by providing automatic unmanned control system with closed circuit television function.

The present invention provides the unmanned automatic control method having the illegal parking and crime prevention closed-circuit television function and system using the vehicle detection method by image analysis. It is effective to eliminate unnecessary system operation due to calculation of unnecessary area and result of image analysis of unnecessary area by limiting to the area of interest rather than setting the whole area of the monitoring area. Secondly, by minimizing the calculation of morphology and labeling of the image analysis by the image blocking method, the real-time detection of the vehicle detection and the success rate of the vehicle detection and tracking can be improved.

In addition, the present invention can be stably realized even in an outdoor environment, and detects the movement of the vehicle in the parking lot monitoring area, tracks it, and detects whether the vehicle is stopped. There is an effect that can selectively process the vehicle that exits and exits, the vehicle to park to avoid shooting.

Hereinafter, with reference to the accompanying drawings, it will be described in detail with respect to the unmanned automatic control system having illegal parking and crime prevention closed circuit television function as a preferred embodiment of the present invention.

1 is an embodiment according to the present invention, Figure 1a is a perspective view schematically showing a state of monitoring and controlling the parking lot monitoring area using an unmanned automatic control system having illegal closed parking and crime prevention closed circuit television function according to the present invention 1B is a block diagram showing a main configuration of a control box, and FIG. 1C is a block diagram showing a network configuration between a plurality of unmanned automatic control systems and a server.

First, the unmanned automatic control system 1A having the illegal parking and crime prevention closed-circuit television function of the present invention is installed around the road 20 so that a plurality of parking monitoring sectors in the parking lot monitoring area 31 of the road 20 are installed. It is a system for monitoring and cracking down on the vehicle 10 illegally parked in the parking lot prohibited area 22 divided by 24). The intermittent control system 1A is equipped with a fixed camera 30 for detection and a P / T (including pan / tilt and zoom function) camera 32 for the detection on an upper part of the support 40 installed on one side of the road 20. On one side of the support (40), a control box (50) capable of automatically cracking down illegal parking is installed. The control box 50 includes the main components of Figure 1b.

The fixed camera 30 for detecting is installed in the parking monitoring area 31 to monitor the parking stop area 22 is divided into the monitoring sector 24 on both sides of the road in proportion to the distance. In addition, the intermittent P / T camera 32 captures an image of a vehicle illegally parked in the prohibited parking area 22 so as to be obtained and utilized as an intermittent material.

The ROI setting module 56 installed in the control box 50 takes a snapshot in real time through the detection fixed camera 30 and takes a snapshot, adds a polygon to the photograph, and adds a polygon to each of the polygons. By pressing and dragging the vertices with a mouse, a plurality of regions of interest (ROI), which are vehicle detection areas for image analysis, are set.

In the present invention, the interest area setting module 56 for introducing a concept of a region of interest (ROI), which is a vehicle detection area, and setting an area of interest easily and quickly and easily for realizing an unmanned automatic parking control system. A / D conversion of the image collected from the fixed camera 30 for the image, and the image transmission module 52 for compressing the image for transmission through the network 60 to the server 70 of the central situation room, and A / D conversion The region of interest is set via the hub 54 with the image of the stopped parking monitoring area 31.

In the setting of the ROI, the scene control is performed by receiving an image of the A / D-converted parking monitoring area 31 and transmitting the image to the vehicle detecting module 59 that detects the parking violation 10 on the parking prohibited area 22. On module 55.

The shadow removal module 58 produces an initial background model with 1 to N images sequentially input from the fixed camera 30 for detection, and produces a difference image between the produced initial background model and the input image after N + 1 th. The foreground region including the moving object and the shadow of the moving object is divided, and the shadow area from the divided foreground area is relatively low in saturation compared to other areas of the foreground area, and the contrast value of the input image and the background image The ratio is included in a certain range even when the environment of crackdown sites is included in a certain range, such as weather changes such as clouds, rain, and snow, altitude changes during the day of the sun, and momentary lighting changes caused by the passing of large vehicles. It is possible to reliably remove the shadow area from the foreground area even if the environment changes.

Application of the shadow removal module 58 can improve the vehicle detection accuracy through the shadow removal that generates inaccurate features in the image analysis, improve the success rate of the vehicle detection, and reduce unnecessary system operation due to the vehicle misdetection.

The adaptive background modeling module 57 removes the moving object region from which the shadow region is removed in the shadow removing module, and takes only the moving object region as the detected moving object region, and the moving object detected in the N + 1th or later input image. Update the background model by including the previously tracked moving object area, except the area.

The vehicle detection module 59 of the present invention divides the image evenly into blocks of a predetermined size in the shadow removing module 58 and performs image blocking to minimize the calculation amount by performing morphology and labeling operations in blocks larger than the pixel unit. To do.

In order for the system using the vehicle detection method by image analysis to be put into practical use in the field, it is important to minimize the massive amount of computation accompanying the image analysis process. The image analysis process after shadow removal consists of expansion and erosion morphology calculations to remove holes inside the partitions and labeling calculations for segmentation. Because morphology and labeling operations are pixel-by-pixel operations of the image, a large amount of computation is required. The computational minimization of image analysis is ultimately achieved by the computational minimization of these two operations. The approach of the present invention is image blocking. The image is evenly divided into blocks of a constant size n × n, and morphology and labeling operations are performed in block units larger than the pixel unit to minimize the computation amount.

In order to realize the unmanned parking control system using the vehicle detection method by image analysis as above, It is possible to minimize the amount of computation in the morphology operation and the labeling operation of the image analysis by the image blocking method.

Vehicle detection module 59 according to the present invention via the image transmission module 52, hub 54, intermittent site control module 55 to the image of the parking lot monitoring area 31 A / D converted in Figure 1a Video block is made by input.

In order for the illegal parking control system using the vehicle detection method by image analysis to be put into practical use in the field, it is important to minimize the massive amount of computation accompanying the image analysis process. The result image up to the shadow removal process is a binary image, and the subsequent image analysis process consists of a morphology calculation of expansion and erosion to remove holes in the partition and a labeling operation for region division. Morphology and labeling operations require large amounts of computation because they are pixel-by-pixel operations of the image. The computational minimization of image analysis is ultimately achieved by the computational minimization of these two operations.

According to the present invention, the image blocking is performed by equally dividing an image into blocks and performing morphology and labeling operations in blocks larger than pixels, which is to minimize the amount of computational analysis in the illegal automatic parking control system of FIG. This can be seen by referring to the flow diagram that shows.

Here, the case where the image size is 320 × 240 pixels and the block size is n = 2 will be described as an example. That is, in step IB1 of the flowchart, a variable ObjCnt for counting pixels of an object and a variable ShadowCnt for counting shadow pixels by the object are used, and are initialized to 0 in step IB1. In the step IB2, the size of the 320 × 240 pixel image is increased by nsize in the horizontal and vertical directions, and a double for loop is performed. The nsize of steps IB2 and IB3 is the width and length of the block. In steps IB3, IB4, and IB6, for each pixel constituting a block of size nsize × nsize, it is determined whether the pixel is an object region or a shadow region pixel, and in steps IB5 and IB7, variables ObjCnt and ShadowCnt are used. To count. Step IB8 determines whether to determine the nsize × nsize block as the area of the object or the area of the non-object. When nsize is 2, as in step IB8, when ObjCnt> 1 or more and ShadowCnt <2, it is determined as an object area. Otherwise, it is determined as a non-object area.

In this way, the size of the calculation is greatly reduced by performing morphology and labeling operations based on the nsize × nsize blocks calculated by the image block.

Therefore, real-time vehicle detection is realized by reducing the morphology calculation amount by image blocking, the labeling calculation amount by image blocking, and the morphology and labeling calculation amount that require a lot of computation time in image analysis, and the realization of vehicle detection is processed. to represent d in an image frame time I _{t- △ t,} the image frame at the current time _t I, is to shorten the time interval △ t increased the success rate of the real-time screen and the vehicle detection and tracking of the vehicle is detected.

Next, referring to the flowchart of the entire system of FIG. 3 and the exemplary diagram of FIG. 4, the parking lot monitoring area 31 in which an image of an enforcement site is taken in real time through a fixed camera 30 for detection installed on a road is described. The region of interest (ROI) is set in the ROI setting module 56 using the input image (CD1, CD2). This is the same as in Figure 4a.

The background model is updated by the adaptive background modeling module 57 (CD3), and the difference operation between the input image and the background image is performed (CD4). This is the same as in Fig. 4b. Then, the shadow removal module 58 removes the shadow area (CD5). This is the same as in FIG. 4C.

After removing the shadow area and modeling the background image, the vehicle detection module 59 blocks the image to minimize the amount of calculation of the image analysis (CD6). After the image blocking, a morphology operation of expansion and erosion for removing holes in the divided region is performed (CD7), and labeling for region division is performed after the morphology operation (CD8). After the labeling operation, a MER (minimum neighboring rectangle) calculation is performed (CD9), each MER representing each divided area after the MER calculation is evaluated to detect a vehicle (CD10), and the vehicle detection information is input. It receives and controls the P / T camera 32 for enforcement to perform illegal parking unmanned enforcement (CD11). This is the same as in FIG. 4D.

While the invention has been shown and described in connection with specific embodiments thereof, it is well known in the art that various modifications and changes can be made without departing from the spirit and scope of the invention as indicated by the claims. Anyone who has a can easily know.

1A is a perspective view schematically showing a state of monitoring and controlling a parking lot monitoring area by using an unmanned automatic control system having an illegal parking and crime prevention closed circuit television function according to the present invention.

Figure 1b is a block diagram showing the main configuration of the control box included in the unmanned automatic control system having illegal closed parking and crime prevention closed circuit television function according to the present invention.

Figure 1c is a block diagram showing the network configuration between the server and the unmanned automatic control system having a plurality of illegal parking and crime prevention closed circuit television function according to the present invention.

FIG. 2 is a flowchart illustrating image blocking for minimizing calculation amount of image analysis in an unmanned automatic control system having illegal parking and crime prevention closed circuit television according to the present invention.

Figure 3 is a flow chart of the overall system in the unmanned automatic control system having illegal parking and crime prevention closed circuit television function according to the present invention.

Figure 4 is an exemplary view according to the operation of the module in the unmanned automatic control system having illegal closed parking and crime prevention closed circuit television function according to the present invention.

♣ Explanation of symbols for the main parts of the drawing ♣

10: vehicle 20: road

22: No parking zone 24: Parking monitoring sector

30: fixed detection camera 31: parking lot monitoring area

32: intermittent P / T camera 40: prop

50: control box 52: video transmission module

54: hub 55: intermittent site control module

56: region of interest setting module 57: adaptive background modeling module

58: shadow removal module 59: vehicle detection module

60: network 70: server

80: printer server 82: printer

1A, 1B, 1C: Unmanned Automatic Control System

Claims (3)

Receiving the image of the intermittent or surveillance site shot in real time through the fixed camera 30 for detection on the road input to the region of interest setting module 56, the region of interest setting module 56 for each major region of interest of the image Illegal parking and crime prevention by image analysis that sets a plurality of regions of interest (ROI), removes shadow areas in the shadow removal module 58, and updates the background model in the adaptive background modeling module 57 In the unmanned automatic control method having a closed circuit television function for (a) The ROI (Region of Interest) is set by performing ROI (Region of Interest) without performing image analysis on the entire image area of the intermittent scene taken in real time through the fixed camera 30 for detection in the entire process of image analysis. Setting a region of interest to limit the region to the target region of the image analysis to achieve a minimum amount of computational analysis; (b) blocking the image in the vehicle detection module 59 in order to minimize the amount of computation of the morphology and labeling in the image analysis process after removing the shadow area and modeling the background image: (c) sequentially performing a morphological operation of expansion and erosion to remove holes in the partition region after the image blocking; (d) performing labeling for region division after the morphology operation; (e) performing a MER (minimum neighboring rectangle) calculation after the labeling operation; (f) detecting the animal body by evaluating each MER representing each area divided after the MER calculation; (g) controlling illegally parked vehicles and performing crime prevention by controlling the P / T camera 32 for receiving and receiving the moving object detection information, the illegal parking and crime prevention closed-circuit television function Unmanned automatic cracking method having a. The method of claim 1, In order to minimize the amount of calculation of the image analysis, Initializing each to 0 using a variable ObjCnt for counting pixels of an object and a variable ShadowCnt for counting shadow pixels by an object, Performing a double for loop by increasing the size of the image (320 × 240 pixels) and nsize (2 × 2 as the horizontal and vertical size of the block) in the row and col directions for the image; Determining whether each pixel constituting a block of size nsize x nsize of the block is a pixel of an object region or a pixel of a shadow region; Digitizing the variables ObjCnt and ShadowCnt in determining the object region and the shadow region pixels; Determining whether to determine the nsize × nsize block as an area of an object or an area of a non-object; If the nsize is 2, ObjCnt> 1 or more, and if ShadowCnt <2, the object area is determined, otherwise the non-object area is determined. Unmanned automatic cracking method having. A fixed camera 30 for detecting a parking stop area 22 installed in the parking lot monitoring area 31 and divided into monitoring sectors 24 on both sides of the road in proportion to the distance; A region of interest setting module 56 for setting a plurality of regions of interest (ROIs), which are vehicle detection regions for image analysis, of an image of a control scene photographed in real time through the detection fixed camera 30; The initial background model is produced from 1 to N images sequentially input from the fixed camera 30 for detection, and the shadow of the moving object and the moving object is a difference image between the produced initial background model and the N + 1th and subsequent input images. A shadow removal module 58 for dividing a foreground area including a shadow and removing a shadow area of a moving object from the foreground area; In the shadow removing module, the moving object area from which the shadow area is removed is taken as the detected moving object area, and the background including the previously tracked moving object area is excluded except for the moving object area detected in the N + 1th input image. An adaptive background modeling module 57 for updating the model; And a vehicle detection module 59 for dividing the image into blocks of a predetermined size in the shadow removal module and performing image blocking to minimize computation by performing morphology and labeling operations in blocks larger than pixels. Unmanned automatic control system with illegal parking and crime prevention closed circuit television features.

Images