KR101001302B1 - A sensing system of a lane offense cars using the corner data and sensing method using the same - Google Patents

Abstract

A first camera for photographing the interruption violation area; A second camera for photographing an interruption violation vehicle detected in the violation area to obtain vehicle information; An image acquisition unit for converting an image captured by the first camera into digital image data; A first image analysis processor configured to analyze the digital image received from the image acquisition unit, extract corner information of the vehicle, and detect an interruption violation vehicle based on the extracted corner information; A second image analysis processing unit which receives the image captured by the second camera and extracts vehicle information of the violation vehicle; A storage unit which stores image information photographed by the first and second cameras; Controlling driving of the first and second cameras based on the image information analyzed by the first and second image analysis processing units, processing the image information analyzed by the first and second image analysis processing units, and storing the image information. Disclosed is a interruption vehicle detection system and a method for controlling a violation vehicle using the same.

Cutting off, violating vehicle, corner information, registration, tracking, detection area

Description

A sensing system of a lane offense cars using the corner data and sensing method using the same}

The present invention relates to a system for detecting an interruption violation vehicle and a method thereof, and more particularly, to a system and a method for detecting an interruption violation vehicle on a road using an image processing algorithm.

As the industry develops, the number of vehicles, the means of transportation, increases, and the traffic volume on the road is steadily increasing. As traffic volume increases, many traffic accidents occur, and many traffic laws are being enforced to prevent such traffic accidents, and one of them is an accident that occurs due to illegally changing lanes on the road. The lane change prohibition line for prevention is marked on the road. Therefore, in the access road of a complicated road, the lane change is prohibited, and the interruption of the intervening vehicle is regulated, leading to the smooth flow of traffic and preventing contact accidents.

On the other hand, various crackdown methods are implemented to crack down on a vehicle changing by changing lanes in a lane change prohibited area as described above, and one of them will be described below.

In other words, there is a method of visually detecting a vehicle to be interrupted while finely intervening traffic police in the area of the violating vehicle control area, and issuing a fine. However, if the traffic police crack down on the traffic police, it is not easy to secure evidence to prove the traffic violation. Therefore, many people have to apply for the traffic violation. There is a problem that must be mobilized. Therefore, where there is no traffic police, there are many violation vehicles that ignore the traffic laws and cause traffic accidents as well as disrupt the traffic flow.

In addition, as another control method, a camera is installed to capture a violation vehicle detection area, and a method of capturing a violation vehicle by capturing the detection area in real time and checking the captured image is used. Since people have to check, there is a problem that requires a lot of manpower and time.

In order to solve these problems, recent attempts have been made to introduce an image detection method using image information processing technology. However, it is necessary to acquire minimum image information for image processing and image processing, which are important limitations in image processing. As the coherence, continuity, and association complexity of information increase, the challenges of exponentially increasing video information throughput and the resulting slowdown in processing speed are solved. In the automatic detection of a violation vehicle, the reliability and processing speed of the processing result are pointed out as a big problem.

SUMMARY OF THE INVENTION The present invention has been made in view of the above, and provides an interruption vehicle detection and enforcement system and method for detecting an interruption vehicle simply and quickly using an image processing algorithm. have.

An interruption vehicle detection system using the corner information of the present invention for achieving the above object comprises: a first camera for photographing an interruption violation vehicle detection area; A second camera for photographing an interruption violation vehicle detected in the violation area to obtain vehicle information; An image acquisition unit for converting the image photographed by the first camera into digital image data; A first image analysis processor configured to analyze the digital image received from the image acquisition unit, extract corner information of the vehicle, and detect an interruption violation vehicle based on the extracted corner information; A second image analysis processing unit which receives the image photographed by the second camera and extracts vehicle information of the violation vehicle; A storage unit which stores image information photographed by the first and second cameras; The driving unit controls the driving of the first and second cameras based on the image information analyzed by the first and second image analysis processing units, and processes the image information analyzed by the first and second image analysis processing units. And a control unit to store in the.

Here, the first image analysis processor sets a normal lane, a detection area located within the normal lane, and a violation predicted lane adjacent to the normal lane in the digital image data, and the normal lane and the violation in the digital image. The corner information of the vehicles located in the predicted lanes may be extracted and registered for each lane, and the registered corner information may be tracked to determine that the vehicle is interrupted when detecting the corner information registered in the predicted lane of violation in the detection area.

In addition, the first image analysis processor compares the ratio of corner information registered in the violation prediction lane among the corner information detected in the detection area with a setting value, and violates when the ratio of corner information exceeds a setting value. It is good to judge that it is a vehicle, and cancel the registration setting of the corner information which passed through the said detection area.

The corner information may include vertices between sides of the vehicles found in the digital image data, and the first image analysis processor extracts the vertices by the brightness difference for each position found in the digital image.

The first image analysis processor may set a determination value for detecting the vertices and register the corner information only when the detected value is equal to or larger than the determination value.

The controller may be further configured to extract image information from a time point at which corner information of the offending vehicle is registered to a time point out of the detection area and store the storing information in the storage unit.

In addition, the interruption violation control method using the corner information of the present invention for achieving the above object, the step of photographing the interruption interruption area of the road; Converting the captured image into a digital image, and setting a normal lane, a violation prediction lane, and a detection area within the normal lane in the digital image; Extracting corner information of the vehicle appearing in the digital image; Registering the extracted corner information for each of the normal lane and the expected violation lane; Detecting the interrupting vehicle by tracking the registered corner information; And acquiring vehicle information by photographing the vehicle detected as the interrupting vehicle.

The extracting may include detecting corner points of a vehicle found in a normal lane and a predicted violation lane for each of the digital image frames; And extracting only corner points that are equal to or greater than a predetermined determination value among the detected corner points.

In the detecting step, the corner point may include a vertex in which the sides or sides of the vehicle meet in the two-dimensional image frame.

The detecting may include: tracking a movement of the registered corner information; Confirming whether corner information registered in the predicted violation lane is found in the normal lane; Setting a danger flag upon detecting corner information registered in a predicted violation lane in the normal lane; Determining whether corner information registered in the anticipated violation lane has entered the detection area while the dangerous flag is set; Calculating the ratio of the corner information amount registered in the predicted violation lane among the corner information found in the detection region when the corner information registered in the predicted violation lane is found in the detection region; And comparing the calculated ratio of corner information with a set value and determining the vehicle as an interruption violation vehicle.

The method may further include: releasing the set flag when the ratio of the calculated corner information amount is less than a set value; And deleting the tracked corner information existing in the detection area.

In the determining step, when the ratio of the corner information registered in the violation prediction lane is 50% or more among all registered corner information detected in the detection area, it may be determined that the vehicle is an interrupted violation.

The method may further include obtaining enforcement information on the vehicle detected as the interrupting vehicle; And storing the obtained intermittent information.

The acquiring may include extracting a vehicle number of the violating vehicle by analyzing the violating vehicle photographed image; Extracting an interrupted image frame included in the vehicle number of the violating vehicle; And extracting the captured image from the point of registration of the violation information to the point of time passing through the detection area.

According to the cornering vehicle detection and enforcement system and method using the corner information of the present invention, it is possible to detect the interrupting vehicle in real time, and process the small amount of data, while more accurate and reliable tracking and detection has the advantage Therefore, there is an advantage that can effectively detect and crack down the interruption vehicle.

With reference to the accompanying drawings it will be described in detail the interruption vehicle detection system and the violation vehicle control method using the corner information according to an embodiment of the present invention.

1, 2, 3a and 3b, the interruption violation detection system using the corner information according to an embodiment of the present invention, the normal lane beyond the interruption prohibition lane 150 of the road 100 It is to detect and crack down on the violation vehicle entering into (110). The interruption violation vehicle detection system includes a first camera 10 for photographing a violation area of the road 100 and a second camera for photographing an interruption vehicle detected in the crackdown area to obtain vehicle information. 20), the image acquisition unit 30 receives the video taken by the first camera 10 and converts it into a digital image, and the vehicle to break the violation by analyzing the digital image received from the image acquisition unit 30 A first image analysis processor 40 for detecting a second image, a second image analysis processor 50 for extracting vehicle information of a violating vehicle by receiving an image captured by the second camera 20, and a storage unit 60 and And a control unit 70 and a transceiver unit 80.

A normal lane 110 is set on one side of the road 100, and a violation predicted lane 120 is set between the normal lane 110 and the cut-off line 150. The first camera 10 photographs the entire road 10 in a predetermined width, and captures a video in real time mainly on the lanes 110 and 120, and the captured video is transmitted to the image acquisition unit 30.

The second camera 20 is installed to photograph a violation vehicle interrupted by the normal lane 110, and the captured image, that is, the still image, is transmitted to the second image analysis processor 50.

The image acquisition unit 30 receives the video captured by the first camera 10, converts the image into digital image data, generates a digital image frame in real time, and transmits the digital image frame to the first image analysis processing unit 40.

The first image analysis processor 40 analyzes the digital image received from the image acquisition unit 30 to extract corner information of the vehicles, registers the extracted corner information for each area (lane), and registers the registered corner information. They will track and detect the interrupting vehicle. In detail, as illustrated in FIG. 2, the first image analysis processor 40 displays corner information in each of the violation prediction lanes 120 adjacent to the normal lane 110 of the road 100 in the digital image 200. Areas 131 and 132 for registration are set. The first image analysis processor 40 sets the detection area 140 at a predetermined width in the normal lane 110, and sets the violation vehicle detection line 141 at the lower end of the detection area 140.

As described above, the corner information of the vehicles 310, 320, and 410 entering the areas 131 and 132 is extracted while the respective areas are set in the digital image 200, and the extracted corner information is registered as lanes 110 and 120. do. The corner information may include color, morphology, size, etc. in addition to two-dimensional corner information and vertex information, which are feature point information included in a target (vehicle) object image. Various conventional techniques, such as those described in Miroslav Trajkovic and Mark Hedley, “Fast corner detection”, Image and Vision Computing, Volume 16, Issue 2, 20 February 1998, Pages 75-87, can be used.

In addition, in order to minimize the distortion caused by the perspective of the image generated due to the physical layout on the road and the characteristics of the camera, a perspective inverse transform image processing algorithm may be applied in the image information preprocessing process.

In the embodiment of the present invention, the corner information of the vehicles 310, 320, 410 is extracted and registered as corner points P1, P2, and P3. That is, the corner points P1, P2, and P3 are registered and managed by the corresponding areas (by lane) for each vehicle 310, 320, and 410.

In the present invention, the corner points P1, P2, and P3 are defined as vertices between planes and planes or sides and sides occurring in the difference of brightness for each position of the vehicle in the two-lane digital image 200. Points that are considered vertices rather than edges are tracked and registered as corner points.

In addition, a specific determination value is set in order to select and register only the corner points to be registered among a large number of corner points to be extracted, and register as a corner point only when the extracted corner point becomes equal to or greater than the set determination value. For example, it is a value obtained using the concept of a so-called CRF (Corner Response Function) as a criterion for calculating a value for the corner point compared with the determination value, and this value may be compared with the determination value. For example, when the corner points are extracted based on the brightness and the like in the image data 200, the determination value may be set based on the brightness of the extracted corner points.

Therefore, if the criterion of the discrimination value is lowered, a large number of corner points are registered, and if the criterion of the discrimination value is increased, there are no corner points at which a few corner points are registered or registered. Therefore, the determination value may be appropriately set in consideration of variable factors such as a location, time (day, night, etc.), season, and weather.

In the same manner as described above, the first image analysis processor 40 extracts and registers corner points P1, P2, and P3 for each vehicle for each frame of the digital image, and registers the registered corner points P1, P2, and P3. Keep track of your movements.

That is, the area having the most similar value is compared with a certain size area around each corner point registered up to the previous video frame and a predetermined size area around each corner point extracted from the current video frame. The coordinates of the registered corner points are moved up to the previous image frame, and the registered corner points can be tracked by repeating this for every image frame. As a method of comparing an area of a predetermined size between the current frame image and the previous frame image, various techniques proposed in the related art may be used. See, eg, K. Briechle, U.D. Hanebeck, "Template Matching using Fast Normalized Cross Correlation", Proc. SPIE, Vol. 4387, 95 (2001), and the like.

The first image analysis processor 40 detects the corner point P3 registered in the violation predicted lane 120 among the tracked corner points P1, P2, and P3 in the normal lane 110, and also detects the detection area. In the case of being tracked and detected at 140, the vehicle may be determined to be an interruption violation vehicle. Here will be described a specific example of determining whether the interruption vehicle. That is, when an object (vehicle) having a specific correlation of the corner points P3 enters the normal lane 110, that is, the normal lane is in the geometrically defined area and reaches the detection area 140. If it does not deviate from (110), it can be determined as a violation vehicle.

More specifically, as shown in FIGS. 3A and 3B, when the corner points P3 to track satisfy the following relation, an object (vehicle) corresponding to the corner points P3 is interrupted. It can be judged to be.

[Relationship 1]

R = B / (A + B)> 50%

Here, A is the number of corner points P3 registered in the violation prediction lane 120 as it is, and B is the number of corner points P3 registered in the violation prediction lane 120. It shows the number of corner points tracked in the normal lane, respectively.

That is, according to Equation 1, when the corner points P3 detected in the vehicle crossing the forbidden line 150 in the violation prediction lane 120, the number of corner points detected in the violation prediction lane 120 is greater than the number of corner points detected in the violation prediction lane 120. When the number of corner points detected in the normal lane 110 is large, it is managed and tracked by the control target vehicle, and when such a state (R> 50%) leads to the detection area 140, it may be determined as an interruption violation vehicle. Will be.

In addition, the corners P1 and P2 are also registered and tracked in the vehicles 310 and 320 driving the normal lane 110, and the movement of the corner points is predicted and tracked for each successive image frame so that the violation prediction lane ( It is possible to track the corner points P3 registered and tracked at 120 and the corner points P1 and P2 while distinguishing from each other. Therefore, even when the vehicle is intermixed with the interruption violation, the process of independently predicting and tracking the corner points P1, P2, and P3 extracted and registered initially is performed. It is possible to detect a violation vehicle interrupted by the normal lane 110. That is, since the structure of the vehicle does not change in a continuous tracking process in the structure of the vehicle, the change in the relative position and number of the corner information P3 hardly occurs, and even if a small change occurs, the specific ratio is higher than a certain ratio. Since it can be determined as the same vehicle, it is possible to track up to the last judgment moment without losing information of the target vehicle registered and tracked from the beginning.

The detection information of the interruption violation vehicle as described above is transmitted to the control unit 70. Here, the first image analysis processing unit 40 is an image from the point of time when the corner information (corner point) of the offending vehicle is registered to the point of leaving the detection line 141 of the detection area 140 when the interrupting vehicle is detected. By extracting the information (video information) to the control unit 70, the control unit 70 can secure the evidence data by storing the received video information in the storage unit 60.

The second image analysis processor 50 receives an image taken by the second camera 20 to obtain the violation vehicle information when the first image analysis processor 40 detects an interruption violation vehicle. Analyze and obtain the vehicle number of the offending vehicle. The obtained information about the violating vehicle (vehicle number, photographed image) is transmitted to the controller 70.

The controller 70 controls driving of the first and second cameras 10 and 20, and detects violation vehicle detection information and evidence information transmitted from the first and second image analysis processing units 40 and 50. The process is stored in the storage unit 60. In addition, the control unit 70 transmits the violation vehicle detection information and evidence information to the remote control center through the transmission / reception unit 80, so that the management center can conveniently and quickly perform the violation vehicle-related duties such as issuing a bill in real time. Manage and process.

A method for cracking down a violating vehicle using the interrupting vehicle detecting and cracking down system according to an embodiment of the present invention having the above configuration will be described.

First, by driving the first camera 10 to capture a violation detection area in real time to obtain a video (S10). The captured video is applied to the image acquisition unit 30 and converted into a digital image (A11).

The digital image (image frame) continuously converted is applied to the first image analysis processor 40, and the first image analysis processor 40 checks whether a detection area is set in the digital image 200 (S12), If not set, the detection area 140 is set as shown in FIG. 2 (S13).

The first image analysis processor 40 extracts corner points P1, P2, and P3 for each vehicle 310, 320, and 410 from the digital image 200 (S14), and extracts the extracted corner points P1, P2, and P3. Each area, that is, lanes 110 and 120 are registered (S15). Here, the method of extracting the corner points P1, P2, and P3 may use the method described above.

Subsequently, the first image analysis processor 40 analyzes the applied digital image 200 to track registered corner points P1, P2, and P3 (S16).

The first image analysis processor 40 checks whether a previously set flag exists (flag value = 1) or not (flag value = 0) (S17).

If the flag is not set in the step S17, it is checked whether there is a violation prediction point (that is, a corner point P3 registered in the violation prediction lane) in the normal vehicle 110 (S18). .

If it is detected that some of the corner points P3 of the vehicle 410 of the violation predicted lane 120 exist in the detection area 140, an interruption violation flag is set (S19).

Subsequently, it is checked whether the corner point P3 is also detected in the violation vehicle detecting line 141 (S20). If detected, the 'R' value is calculated by the relational expression 1 above (S21).

And if it is determined that R> 50% (S22), and satisfies the relation 1, it can generate an alarm for the interruption violation (S24). The alarm information is applied to the controller 70, and the controller 70 controls the second camera 20 to photograph the violating vehicle 410 (S25).

The second image analysis processor 50 obtains an image of the violating vehicle and obtains the vehicle number of the violating vehicle from the obtained image (S26).

The controller 70 stores the video information of the road and the captured image (still image) of the violation vehicle, respectively, in the storage unit 60 (S27).

In addition, the information obtained in the step (S26) may be transmitted to the management center of the remote place through the transmission / reception unit 80 (S28).

On the other hand, if it is determined in step S22 that the relation 1 is not satisfied, the set interrupt danger flag is released, and the tracked corner point existing on the vehicle entering the detection line 141 is deleted (S23). .

By the above method, the vehicles violating the interruption on the road can be detected quickly in real time, and in particular, it is possible to detect more accurately and quickly while being less affected by the surrounding environment.

In addition, by extracting and tracking only a desired amount of corner points, the amount of data to be processed can be reduced, so that real-time detection of interruption violations occurring on the road can be performed and the reliability of the detection can be improved.

1 is a schematic configuration diagram showing an interruption vehicle control system according to an embodiment of the present invention.

2 is a view for explaining an image frame photographed by the first camera of FIG.

3A and 3B are diagrams for explaining a method of detecting a violation vehicle by tracking corner points for each vehicle on an image frame.

Figure 4 is a flow chart illustrating a interruption vehicle control method according to an embodiment of the present invention.

<Description of Symbols for Main Parts of Drawings>

10. First camera 20. Second camera

30. Image acquisition unit 40. First image analysis processing unit

50. Second image analysis processing unit 60. Storage unit

70. Control unit 80. Transmitter / receiver unit

Claims (14)

A first camera for photographing an interruption vehicle detection area; A second camera for photographing an interruption violation vehicle detected in the violation area to obtain vehicle information; An image acquisition unit for converting the image photographed by the first camera into digital image data; A first image analysis processor configured to analyze the digital image received from the image acquisition unit, extract corner information of the vehicle, and detect an interruption violation vehicle based on the extracted corner information; A second image analysis processing unit which receives the image photographed by the second camera and extracts vehicle information of the violation vehicle; A storage unit which stores image information photographed by the first and second cameras; The driving unit controls the driving of the first and second cameras based on the image information analyzed by the first and second image analysis processing units, and processes the image information analyzed by the first and second image analysis processing units. Cut-off violation vehicle detection system using the corner information, characterized in that it comprises a control unit for storing. The method of claim 1, wherein the first image analysis processing unit, Setting a normal lane, a detection area located within the normal lane, and a violation prediction lane adjacent to the normal lane in the digital image data; Extract corner information of the normal lane and the vehicles located in the violation predicted lane from the digital image and register the corner information for each lane, track the registered corner information, and detect the corner information registered in the detection predicted lane in the detection area. Interrupting violation detection system using the corner information, characterized in that it is determined to be a city interruption vehicle. The method of claim 2, wherein the first image analysis processing unit, Among the corner information detected in the detection area, the ratio of the corner information registered in the violation prediction lane is compared with a setting value, and when the ratio of the corner information exceeds the setting value, it is determined as a violation vehicle and passes through the detection area. An interruption violation detection system using corner information, characterized in that to cancel the registration setting of one corner information. The method of claim 2, wherein the corner information includes vertices between sides of vehicles found in digital image data. The first image analysis processor extracts the vertex based on the brightness difference for each position found in the digital image. The method of claim 4, wherein the first image analysis processor sets a discrimination value for detecting the vertices and registers corner information only when the detected value is greater than or equal to the discrimination value. Violation Vehicle Detection System. The method according to any one of claims 2 to 5, wherein the control unit, When the interruption violation vehicle is detected, the detection of the interruption violation vehicle using the corner information comprises extracting and storing the image information from the time point at which the corner information of the violation vehicle is registered to the point out of the detection area and storing it in the storage unit. system. Photographing the interruption detection area of the road; Converting the captured image into a digital image, and setting a normal lane, a violation prediction lane, and a detection area within the normal lane in the digital image; Extracting corner information of the vehicle appearing in the digital image; Registering the extracted corner information for each of the normal lane and the expected violation lane; Detecting the interruption violation vehicle by tracking the registered corner information; And And obtaining vehicle information by photographing the vehicle detected as the interrupting vehicle. The method of claim 7, wherein the extracting step, Detecting corner points of a vehicle found in a normal lane and a predicted violation lane for each of the digital image frames; And extracting only corner points that are equal to or greater than a predetermined determination value among the detected corner points. The vehicle according to claim 8, wherein in the detecting step, the corner point comprises a vertex in which a side or a side and a side meet between sides of the vehicle in a 2D image frame. Detection method. The method of claim 7, wherein the detecting step, Tracking the movement of the registered corner information; Confirming whether corner information registered in the predicted violation lane is found in the normal lane; Setting a danger flag upon detecting corner information registered in a predicted violation lane in the normal lane; Determining whether corner information registered in the anticipated violation lane has entered the detection area while the dangerous flag is set; Calculating the ratio of the corner information amount registered in the predicted violation lane among the corner information found in the detection region when the corner information registered in the predicted violation lane is found in the detection region; And comparing the calculated corner information amount with a set value to determine the interruption violation vehicle. The method of claim 10, further comprising: releasing the set flag when the calculated ratio of the corner information amount is less than a set value; And And deleting the tracked corner information existing in the detection area. The method of claim 10, wherein in the determining step, The interruption violation detection method using the corner information, characterized in that it is determined as an interruption violation vehicle when the ratio of the corner information registered in the violation prediction lane is 50% or more among all registered corner information detected in the detection area. The method according to any one of claims 7 to 12, Obtaining enforcement information on the vehicle detected as the interrupting vehicle; And The interruption violation detection method using the corner information, characterized in that it further comprises the step of storing the obtained enforcement information. The method of claim 13, wherein the obtaining step, Extracting a vehicle number of the violating vehicle by analyzing the violating vehicle photographed image; Extracting an interrupted image frame included in the vehicle number of the violating vehicle; And extracting a photographed image of the violation vehicle from the time point of registering the corner information to the time point of passing through the detection area.

Images