KR100706599B1 - Parking/stopping vehicles detection system and method - Google Patents

Abstract

The present invention is a network camera is installed in the prohibited parking area, the IP is set to collect the image of the area; A network for transmitting the compressed image data collected through the camera to a management server; A video restoring unit for restoring an image from a video compressed bit stream of the network camera transmitted through the network; A sum image preprocessing unit configured to generate a sum image by performing an average algorithm when the successive predetermined frame images restored by the moving image restoration unit are transmitted; An edge detector configured to transmit an aggregated image from the aggregated image preprocessor, to perform edge enhancement filter processing on the aggregated image, and to detect an edge having a predetermined threshold value or more, and to generate an edge image; An edge image storage unit storing the edge image detected by the edge detector for a predetermined time; And a vehicle edge detector for comparing and analyzing the edge image transmitted from the image storage unit, in order to monitor a vehicle illegally parked or stopped in a parking prohibited area, without requiring a separate parking violation monitoring agent. Can automatically detect parking stops.

Description

Parking / Stopping Vehicles Detection System and Method

1 is a block diagram of a parking vehicle detection system according to the prior art,

Figure 2a is a block diagram of a parking vehicle detection system according to the present invention,

2b is a detailed configuration diagram of a vehicle edge detector according to the present invention;

3a to 3b is an operation flowchart of a parking vehicle system according to the invention,

4A to 4C are exemplary views of edge images extracted according to the operation sequence of the parking vehicle system according to the present invention.

5 is an exemplary view of summed image preprocessing according to the present invention;

6a to 6b are exemplary views showing the result of a summation of a parked vehicle prohibited area according to the present invention;

7 is an exemplary edge detection graph shown in accordance with the road situation according to the present invention.

* Description of the symbols for the main parts of the drawings *

20: camera 30: network

40: video control module 41: video restoring unit

42: sum image preprocessing unit 43: edge detection unit

44: edge image storage unit 45: vehicle edge detection unit

50: control unit

The present invention relates to a parking vehicle detection system, and more particularly, when a vehicle remotely parked or parked in an illegal parking prohibition zone via the Internet receives a video of the process and automatically detects the parking vehicle through an image analysis algorithm. And a parking vehicle monitoring system capable of performing the same.

Increasing industrial society and economic growth have led to an increase in the number of automobiles and driving populations. Accordingly, illegal parking and parking on heavy traffic roads have enormous obstacles to normal traffic flows. It is necessary to crack down on illegal parking and stopping vehicles.

There are two types of crackdowns on illegal parked vehicles: one is the artificial crackdown by illegal parking crackers and the other is a remote crackdown using surveillance cameras.

The crackdown of artificial illegal parking vehicles is made through direct sticker issuance and photographing by public service personnel, local government officials, medical officers, etc. Each camera is installed with a remote monitoring control, and through this, the management server detects illegal parking vehicles in the parking prohibited area.

In case of illegal parking management using passive method by enforcement officers, the parking capacity is absolutely insufficient compared to the number of vehicles in Korea at present, and in reality, manual illegal parking control by enforcement officers is carried out from outside and caught. Books, cameras, stickers, such as a variety of crackdown related tools have to work on the crackdown, so the inconvenience is a lot of time due to this situation is bound to delay. In addition, the operation was inefficient due to the hassle of returning to the ward office after the crackdown and re-enter the data on the caught vehicle.

1 is a block diagram of a general illegal parking vehicle automatic detection system, a fixed camera 11 for monitoring the illegal parking forbidden area on the road and photographing a vehicle entering the prohibited area, tracking recognition of the vehicle entering the forbidden area Speed dome camera 12 with a pan / tilt zoom function to capture and collect vehicle information, a motion detection module, and a CPU 16 for controlling the operation of the fixed camera and the speed dome camera, the fixed camera, and the speed dome. The control module 13 for controlling the camera, the image module 14 for processing the captured image signal of the fixed camera and the speed dome camera, and the motion detection module 15 for providing a parking stop vehicle detection function through the object extraction to the CPU It is.

The operation of the system configured as described above is as follows.

When a change in motion occurs in the captured image by entering or exiting a vehicle forbidden zone monitored by the fixed camera 11, the image is compressed through a motion vector extracted from the MPEG compression region by the motion detection module 15. Object segmentation and extraction are carried out to determine whether the vehicle has entered or exited a parking vehicle.

When the captured image of the fixed camera 11 is processed and the vehicle enters the prohibited parking area, the captured image by the fixed camera is provided and processed through the image module 14 and the motion detection module 15 is processed. The control command of the CPU 16 is transmitted to the fixed camera 11 again through the control module 13 to take a picture of entering the parking stop area of the violation vehicle.

The problem is that the vehicle is far from the camera (about 50m or more), the size of the vehicle is smaller than the size of the macroblock, the motion vector does not occur. Therefore, there is no method for effectively detecting a stop of a long distance vehicle.

In addition, when the vehicle is moving very slowly, the motion vector does not appear, so only the edge value is changed in the difference image between frames, and there is no motion, and a fatal defect appears. There are many.

The present invention was created in view of the above conditions, and an object of the present invention is to receive an image of a vehicle parked or stopped in an illegal parking prohibited area collected by a network camera via the Internet, and to receive an image from a remote management server. By automatically detecting illegal parking lots, it can reduce unmanned personnel by facilitating unmanned parking lot, reduce the administrative costs, process quickly and accurately, and detect parking vehicles that can be detected automatically even if the size of the vehicle image is small. A system and method are provided.

Another object of the present invention is to provide a parking vehicle detection system and method having stability which is hardly influenced by external noise by applying summation preprocessing after restoring video data of an illegal vehicle transmitted through the Internet.

Still another object of the present invention is to provide a reliable parking vehicle detection system and method having high reliability since it is possible to minimize the false detection rate according to the situation of the road by utilizing information of about 200 preprocessed edge images.

Technical means according to the present invention for achieving the above object is a network camera is installed in the prohibited parking area, IP is set to collect the image of the area; A network for transmitting the compressed image data collected through the camera to a management server; A video restoring unit for restoring an image from a video compressed bit stream of the network camera transmitted through the network; A sum image preprocessing unit configured to generate a sum image by performing an average algorithm when the successive predetermined frame images restored by the moving image restoration unit are transmitted; An edge detector configured to transmit an aggregated image from the aggregated image preprocessor, to perform edge enhancement filter processing on the aggregated image, and to detect an edge having a predetermined threshold value or more, and to generate an edge image; An edge image storage unit storing the edge image detected by the edge detector for a predetermined time; And a vehicle edge detector which compares and analyzes the edge image transmitted from the image storage unit and determines whether the edge is the edge of the vehicle or the edge of another object.

In order to achieve the above object, the technical method according to the present invention includes a method for analyzing an image of a parking prohibited area, the method comprising: collecting an image of the parking prohibited area through the camera and transmitting the image to the image control module; Restoring the compressed video after receiving the compressed video through the camera; Generating a summed image of a predetermined number of consecutive frame images among the reconstructed images through an average algorithm; Generating and storing edge images after detecting edges above a predetermined threshold by performing edge enhancement filtering on the summed images; And detecting an illegal parking vehicle by detecting an edge of the parking vehicle in the stored edge image.

According to the present invention, in monitoring a vehicle that is illegally parked or parked in a prohibited area of a state or stop, there is no need for a separate parking violation monitoring agent to quickly and accurately crack down on illegal parking, thereby increasing costs. I can solve it.

In addition, after reconstructing the image data received from the Internet all the time, a summing preprocessor is applied, which is more stable than other conventional vehicle automatic detection systems using simple two-dimensional image pixel values.

In addition, by using information of about 200 pre-processed edge images in time rather than a simple edge detection algorithm, the reliability is very high while minimizing the false detection rate according to the road situation.

In addition, vehicles located at long distances can monitor vehicles smaller than the macro block size of video compression.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings.

2A is a block diagram of a parking vehicle detecting system according to the present invention, which is composed of a network camera 20, a network 30, an image control module 40, and a controller 50, and the image control module 40 is a video. The decompression unit 41, the sum image preprocessing unit 42, the edge detection unit 43, the edge image storage unit 44 and the vehicle edge detection unit 45.

The camera 20 is a MPEG 4 fixed network camera capable of compressing video. The camera 20 is installed above a prop (not shown) for observing a parking prohibited area at a certain height, and collects a video of a vehicle parked in each illegal parking prohibited area. Compress and send to the remote management server.

In addition, the camera 20 is assigned a floating IP by an illegal parking vehicle management server (not shown), and the plurality of network cameras installed in each area have different floating IPs assigned by the management server.

That is, the control unit (not shown) of the network camera compresses the collected image by MPEG-4 method, and when the request for image transmission is received from the control unit 50 of the remote management server, the compressed image is compressed by the MPEG-4 method. Will be sent over the network.

By using an IP camera, the initial installation cost is the lowest, maintenance-free, and highly scalable product that enables real-time monitoring of remote locations through the internet 24 hours a day, 365 days a year.

In addition, firmware can be upgraded via the Internet, monitored by an existing PC, and coped remotely in case of emergency.

The network 30 is a data communication system that transmits compressed image data collected through a network camera to a management server and transmits an instruction command signal of the management server to the network camera. The network can combine or selectively use all wired and wireless, including wireless LAN, ADSL, leased line, and the like.

The controller 50 instructs the operation of the network camera 20, assigns a floating IP to the plurality of network cameras, and instructs the operation of the image control module 40, but is input through an input device (not shown). The network camera and the image control module are controlled to perform a command corresponding to a user's arbitrary indication signal.

That is, the network camera 20 may be instructed to upgrade, divide monitoring, emergency response, and the like, and set a specific period to search for illegal parking vehicles caught in the corresponding period.

In addition, since the connection between the network camera 20 and the management server for image reception is possible through a wired Internet network of TCP / IP protocol, a wireless Internet network such as WAP protocol, and a short range wireless relay network through separate wireless equipment, The management server connected to the network camera 20 having a built-in web server can be connected to the network camera only by inputting an Internet address for the network camera into a web browser, thereby obtaining control of the network camera. As a result, real-time monitoring data reception and camera function control can be performed.

The video restoring unit 41 restores the video from the video compressed bit stream transmitted from the MPEG 4 fixed network camera 20 installed remotely over the Internet.

That is, a codec for restoring the MPEG-4 stream is inserted in the program set in the image control module of the remote management server. The MPEG received by requesting the image collected from the camera by connecting to the network camera according to the instruction of the controller -4 Restore video from stream.

The summed image preprocessing unit 42 receives the 64 consecutive frame images in time from the video restoring unit 41 and applies them to Equation (1) below to perform average. It will be described in the operation description with reference to FIG.

-식(1)

Formula (1)

When the averaged preprocessing is performed on 64 frames of images through the summed image preprocessing unit 42, the phenomenon as shown in FIGS. 6A and 6B is displayed.

FIG. 6A is a summed image preprocessing image of a vehicle illegally parked in a parking prohibited area, and the image inside the rectangle in FIG. 6B is a summed image preprocessing image of a vehicle moving in a parked prohibited area. In this case, the stopped vehicle is clearly displayed even if it is averaged, and a summed image having a characteristic of blurring the moving object or the vehicle can be obtained, and the influence of noise can be reduced by the average.

In addition, the preprocessing process greatly contributes to the removal of noise that may occur during image acquisition in the video.

The edge detector 43 receives the summed image summed up by the summed image preprocessing unit 42 and performs edge enhancement filter processing on the summed image to find edges having a specific threshold value or more, and generates an edge image to generate the edge image storage unit 44. To send.

The edge image storing unit 44 stores the edge image generated by the edge detecting unit 43 at a predetermined time interval, and extracts the corresponding edge image stored according to a command transmitted from the vehicle edge detecting unit 45. 45).

In addition, the edge image storage unit 44 stores edge images of about 200 frames in time. By using this, the edge image of the parking line, shadow, etc., which always appear on the road, is stored to be compared with the edge of the current parking vehicle image, and transmitted to the vehicle edge detector 45.

The vehicle edge detector 45 includes a primary edge detector 45-1, a secondary region detector 45-2, and a tertiary edge length detector 45-3, and is transmitted from the edge image storage unit 44. It is determined whether the edge image is an edge image of the vehicle or an edge image of another object except the vehicle.

In addition, the vehicle edge detector 45 may receive an edge image of about 200 frames from the edge image storage unit 44 and determine whether the edge image is a moving vehicle edge image in consideration of their correlation characteristics.

Since the same location edge is displayed continuously in the shadow or parking line that appears on the road, such as a parked vehicle, it is determined whether the edge existed at the same location before or when the edge was created through the vehicle edge detector. To judge.

4A is an edge image generated by the edge detecting unit 43 on the road, in which all the parking line of the road, the shadow of the building, the stopped vehicle already stopped and the edge of the vehicle just stopped are shown.

FIG. 4B illustrates a result of detecting a parking vehicle edge after comparing the edge of the current image and the edge of the preceding image by the vehicle edge detector, and FIG. 4C illustrates an image of a vehicle currently illegally parked according to the result of FIG. 4B. will be.

The primary edge detector 45-1, the secondary flat region detector 45-2, and the tertiary edge length detector 45-3 of the vehicle edge detector 45 will be described with reference to FIGS. 2B and 7. .

The primary edge detector 45-1 compares the edge value of one edge image transmitted from the edge image storage unit 44 with the first threshold value TH1, and transmits the edge image storage unit 44. If the edge value of one edge image is larger than the predetermined first threshold value TH1, it is determined as an image candidate of the vehicle edge, and is transmitted to the secondary region detector 45-2.

At this time, since the shadow area of the fixed lane, pedestrian crossing, road, etc. of the road also has an edge, it passes through the primary edge detector 45-1 and is transmitted to the secondary area detector 45-2.

The secondary region detector 45-2 compares and analyzes the edge position of the image transmitted from the primary edge detector 45-1 with the edge position of the preceding image, and determines that the vehicle is the primary edge detector 45-1. Comparing and analyzing the edge value of the preceding image of the same position and the second threshold (TH2), the edge value of the preceding image of the same position as the edge determined to be the vehicle in the primary edge detector 45-1 If it is less than or equal to the threshold value TH2, it is determined that the vehicle is transmitted to the third edge length detector 45-3.

At this time, the secondary region detector 45-2 compares and analyzes edge values between previous 101 to 200 frames with respect to edge positions of the image transmitted from the primary edge detector 45-1.

At this time, the edge of the fixed lanes and building shadows, which are not the vehicle, is filtered because the edge value does not decrease.

The third edge length detector 45-3 determines whether the edge passing through the first edge detector 45-1 and the second region detector 45-2 has disappeared within the time of the third threshold TH3, If the edge passing through the primary edge detector 45-1 and the secondary region detector 45-2 is maintained without disappearing the edge for more than a predetermined time of the third threshold TH3, it is determined that it is a parking vehicle, not a noise. The vehicle parking stop detection result is transmitted to the controller 50.

In addition, it is compared whether the length of the edge image determined as the stop vehicle is too large or too small for the size of the specific vehicle.

3A is an operation configuration diagram of a parking lot detection system, and the illegal vehicle detection method of the parking vehicle detection system will be described below.

First, the image of the parking stop area photographed by the network camera 20 is acquired through the network (S1), and then the image restoration unit 41 restores the image from the video compression bit stream (S2), and periodically transmits the continuous As shown in FIG. 5, the summed images are added to the time axis by 64 frames, and then the summed images are preprocessed (S3).

If the compressed image of the parking prohibited area photographed by the camera 20 is transmitted from a remote place through a network to detect the parking vehicle directly from the parking prohibited area, instead of detecting the parking vehicle, the parking stop is prohibited. By directly connecting and connecting a terminal (not shown) equipped with an image control module and a control unit to the camera installed in the area, the image of the camera can be directly transmitted through the terminal in the field.

The terminal is a terminal connected directly to the camera, and the image captured by the camera is transmitted to the image control module 40 as a general image, not a compressed image.

Therefore, since the image control module 40 does not need the image restoring unit 41, the terminal includes the combined image preprocessing unit 42, the edge sensing unit 43, the edge image storing unit 44, and the vehicle edge detecting unit 45. It is composed of

Accordingly, the image of the parking stop prohibition area acquired through the camera (S1) is transmitted to the sum image preprocessing unit 42 of the image control module. After summing up, the summed images are preprocessed (S3).

That is, when the summed image preprocessing unit 42 performs preprocessing of 64 frames of images through an average algorithm, the stationary vehicle and the moving vehicle may be clearly distinguished from each other, as shown by a box of FIG. 6A. do.

That is, the image of the stationary vehicle summed up by the sum image preprocessing unit 42 has the edge of the vehicle itself because the image is clear even if the average is applied, and the edge of the vehicle is not fixed because the moving movement is performed. Will be blurred.

Using the edge features of the fixed object, the parking vehicle, and the moving vehicle in the summed image, the edge detector 25 detects an edge of the edge of the image preprocessed with the summed image above a certain threshold, as shown in FIG. 4A. Create and save (S4).

The edge image storage unit 44 stores the current edge image, and also stores the edge value and the edge image of each object according to the change of time shown in FIG. When the vehicle edge is detected, the control unit 50 or the vehicle edge detector 45 transmits the vehicle edge detector 45 according to the instruction.

An operation process of detecting a parking vehicle by analyzing the image collected by the camera will be described in detail with reference to FIG. 7.

The graph of FIG. 7 is an edge generation graph of the summed image according to various conditions of the road, and graph a shows edge values of shadows or parking line edges which are still in the parking stop area according to the change of time. Graph d shows the edge value of the edge when the vehicle passes a shadow or parking line and the edge blurs for a while and then becomes sharp again. Graph b shows the current time (t) of an edge that has not existed before. An edge is generated at the edge of the vehicle that is currently parked and the edge value of the vehicle that is currently parked according to the change of time.

The vehicle edge detector 45 compares and analyzes the edge image stored in the edge image storage unit 44 to detect a parking stop of the vehicle as shown in FIG. 4B (S5), and the image of the parking stop detecting vehicle as shown in FIG. 4C or the vehicle parked correctly. The image of the part and its position information are transmitted to the controller 50.

The vehicle edge detector 45 detects the parking stop of the vehicle by using the third-order detailed detector as described below with reference to FIGS. 3B and 7.

Among the edge images stored in the edge image storage unit 44, the edge value of the current edge image is compared with the first threshold value TH1 (S51), and when it is larger than the first threshold value TH1, it is determined as a vehicle edge (S52). At this time, since the shadow area of the fixed lane or pedestrian crossing, the building, etc. of the road also has an edge, it passes through the primary edge detector 45-1.

Thereafter, the edge position passing through the primary edge detector 45-1 and the edge position of the preceding 101-200 frame are compared (S53), and then the edge value between the preceding 101-200 frames among the edges of the same position is always the second threshold. If less than (TH2) (S54) it is determined as the vehicle edge (S55). At this time, the edge of the fixed lanes and building shadows, which are not the vehicle, is filtered because the edge value does not decrease.

After that, when the edge passing through the primary edge detector 45-1 and the secondary region detector 45-2 is maintained without the edge disappearing for a predetermined time of the threshold TH3, the tertiary edge length detector is maintained (S56). It is determined that the illegal parking vehicle is not a noise (S57) and transmits an image of the parking vehicle to the controller 50. In addition, it compares whether the edge length of the vehicle determined to be an illegal parking vehicle is not too large or too small for the size of a specific vehicle.

The present invention is not limited to the above-described embodiments, and various changes can be made by those skilled in the art without departing from the gist of the present invention as claimed in the following claims.

As described in detail above, the present invention can be controlled quickly and accurately with minimal manpower, since it is possible to crack down without having to wait for a separate parking violation monitoring agent at all times in order to monitor a vehicle illegally parked or stopped in a parking prohibited area. It is possible to solve the problem of increasing the costs and thus increase the cost, and even if two or more vehicles simultaneously violate the parking in the designated parking prohibited area, it is possible to handle multiple control.

In addition, by reconstructing the image of the video data received from the Internet and applying the summing preprocessor, it is possible to have stability that is not affected by external noise, compared to other conventional vehicle automatic detection systems using simple two-dimensional image pixel values. By using the Internet to crack down on illegal parking, you can shoot a video around the clock for 24 hours and send it to a management server to monitor the video at a remote monitoring center in real time to crack down on illegal parking violations. It allows you.

In addition, by using the information of about 200 pre-processed edge images in time rather than a simple edge detection algorithm, the reliability is very high while minimizing the false detection rate according to the road situation.

In addition, compared to the method using a motion vector, a fixed camera can detect a parking stop violation of a vehicle even without a distant distance of more than 50m.

Claims (6)

A network camera installed in a prohibited parking area and configured to collect an image of a corresponding area; A network for transmitting the compressed image data collected through the camera to a management server; A video restoring unit for restoring an image from a video compressed bit stream of the network camera transmitted through the network; A sum image preprocessing unit configured to generate a sum image by performing an average algorithm when the successive predetermined frame images restored by the moving image restoration unit are transmitted; An edge detector configured to transmit an aggregated image from the aggregated image preprocessor, to perform edge enhancement filter processing on the aggregated image, and to detect an edge having a predetermined threshold value or more, and to generate an edge image; An edge image storage unit storing the edge image detected by the edge detector for a predetermined time; And And a vehicle edge detector for comparing and analyzing the edge image transmitted from the image storage unit with a preceding edge image. A camera installed in the prohibition stop area and collecting an image of a corresponding area; A summed image preprocessor connected to the camera and configured to generate a summed image by performing an average algorithm when the continuous predetermined frame image collected by the camera is transmitted; An edge detector configured to transmit an aggregated image from the aggregated image preprocessor, to perform edge enhancement filter processing on the aggregated image, and to detect an edge having a predetermined threshold value or more, and to generate an edge image; An edge image storage unit storing the edge image detected by the edge detector for a predetermined time; And And a vehicle edge detector for comparing and analyzing the edge image transmitted from the image storage unit with a preceding edge image. The vehicle edge detector of claim 1 or 2, wherein the vehicle edge detection unit A primary edge detector for determining a parking vehicle by comparing an edge value of the edge image transmitted from the edge storage unit with a first threshold value TH1; The secondary edge detector which compares the leading edge and the edge of the specific position determined as the parking vehicle in the primary edge detector, and compares the leading edge value and the second door deceleration TH2 among the edges of the same position to determine the parking vehicle. ; And If the edge image determined as the vehicle is maintained for more than a predetermined time of the third threshold (TH3) by the secondary region detector, it is determined as a vehicle, and the third order comparing the length of the edge image determined as the stopped vehicle and the size of a specific vehicle Parked vehicle detection system comprising an edge length detector. In the method of analyzing the image of the no parking zone, Collecting an image of the parking stop area through a camera and transmitting the image to the control unit of the management server; Restoring the compressed video in the video restoring unit after receiving the compressed video through the camera; Generating, by the sum image processing unit, a sum image of a predetermined number of frame images among the reconstructed images through an average algorithm; Generating and storing an edge image after detecting an edge having a predetermined threshold value by performing edge enhancement filter processing on the summed image; And And detecting an illegal parking vehicle by detecting an edge of the parking vehicle in the edge image stored by the vehicle edge detector. delete The method of claim 4, wherein the detecting of the edge of the parking vehicle in the stored edge image comprises: Determining an edge of a vehicle when an edge value of an edge image transmitted from the edge image storage unit to a primary image detector is equal to or larger than a first threshold value TH1; Determining the vehicle edge when the vehicle edge determined by the primary image detector is less than or equal to a predetermined second threshold value TH2 for a predetermined time; And And determining that the vehicle stops when the edge of the vehicle determined by the secondary region detector is maintained for more than a predetermined time of the third threshold (TH3).

Images