TWI808603B - A ramp exit road condition detection system, method, and computer-readable medium thereof - Google Patents

Abstract

The present invention provides a ramp exit road condition detection system, method and computer-readable medium thereof, including a first photographing device, a second photographing device, and a road condition detecting device, wherein the first photographing device records a first image of a target area and at least one target vehicle in the target area, and the road condition detection device uses the first image to track vehicle driving. When the target vehicle drives in violation, a violation identification result is generated, and the second camera device is caused to record a second image for evidence search. In this way, the present invention can instantly detect the target vehicle driving illegally, restore the current driving situation of the target vehicle, and quickly and precisely provide evidence to law enforcement agencies for further processing on the illegal vehicle.

Description

A ramp exit road condition detection system, method and computer-readable medium thereof

The present invention relates to a road condition detection technology, in particular to a ramp exit road condition detection system, method and computer-readable medium thereof.

In today's society, due to the convenience of transportation brought by expressways, many people will use expressways to drive to and from various counties and cities. However, illegal driving incidents are also prone to occur, for example, illegal driving incidents such as drivers crossing grooved lines, driving on road shoulders, and arbitrarily switching lanes due to negligence.

However, the existing technology can only record the information of violating vehicles through monitoring equipment, such as license plates, etc., but cannot effectively monitor the driving conditions of the vehicles, and requires a lot of manpower to maintain the operation of the monitoring system, and cannot automatically judge whether the vehicles have illegal driving events in real time. Therefore, the existing technology not only makes the monitoring cost higher, but also ignores potential risks due to human negligence.

Therefore, how to provide an on-ramp exit road condition detection mechanism to effectively monitor vehicle driving conditions and determine whether vehicles have illegal driving events in an instant, fast and accurate manner has become an urgent issue to be solved in the industry.

In order to solve the technical problems of the aforementioned learning or provide relevant effects, the invention provides a ramp export road condition detection system, which includes: first photography device, the first image of the target area of the shooting ramp exit, and at least one target vehicle in the target area; the detection device is connected to the first photography device to receive the first image and use the first image to perform a one. After the tracking processing, the vehicle tracked by the target vehicle driving in the target area in the first image, which confirms that the target vehicle has produced an illegal identification result when driving in the target area; and a second photography device, which is a communication device that connects the road conditions to the road conditions detection device. In this target area, the second image of the target vehicle is obtained in accordance with the second image to obtain the illegal certificate of the target vehicle, and then transmitted the illegal certificate to the road conditions detection device. The second image of the target vehicle, the illegal evidence, and the results of the violation of the rules are an illegal driving information.

The present invention further provides a road condition detection method at a ramp exit, which includes: taking a first image of a target area at the ramp exit and at least one target vehicle in the target area by a first photographing device; receiving the first image by the road condition detection device, using the first image to perform object tracking processing, and then performing vehicle tracking on the target vehicle driving in the target area in the first image, so that when it is confirmed that the target vehicle is driving illegally in the target area, a violation identification result is generated; and when the road condition detection device generates the violation When the identification result of the regulation is obtained, a second photographing device captures a second image of the target area and the target vehicle in the target area to obtain the violation evidence of the target vehicle based on the second image, and then transmits the violation evidence to the road condition detection device so as to integrate the second image of the target vehicle, the violation evidence and the violation identification result into a violation driving information.

In one embodiment, the target area is a shoulder, an off-ramp road, a grooved line or a general road.

In one embodiment, the road condition detection device includes an image recognition module, wherein the image recognition module includes: a grooved line recognition unit that performs the object tracking process based on the first image to identify whether the target vehicle crosses the grooved line in the target area; a road shoulder recognition unit that performs the object tracking process based on the first image to identify whether the target vehicle is driving on the road shoulder in the target area during the detection time period; and a lane recognition unit that performs the object tracking process based on the first image. Object tracking processing to identify whether the target vehicle is on an off-ramp road or a switchable lane on a general road in the target area for lane switching, wherein the image recognition module generates the violation identification result when the target vehicle crosses a grooved line in the target area, the target vehicle travels on a shoulder in the target area during the detection time period, or the target vehicle is not located in a switchable lane on an off-ramp road or a general road in the target area.

In one embodiment, when the lane recognition unit recognizes that the target vehicle is located on an off-ramp road or a normal road in the target area to switch lanes, the lane recognition unit performs perspective transformation on the actual lane picture of the off-ramp road or normal road in the target area in the first image to transform into a new lane picture, and then obtains the positions of the target vehicle and other vehicles, and the lane recognition unit calculates whether a preset distance is maintained between the target vehicle and the other vehicles, so that when the preset distance is not maintained, the The image recognition module generates the violation recognition result.

In one embodiment, the road condition detection device includes a violation identification module for receiving the violation recognition result so that the second photographing device can capture the second image of the target vehicle so as to display the violation driving information on a screen in real time.

The present invention also provides a computer-readable medium, which is applied to a computer or computing device with a processor and/or memory, the computer or the computing device executes a target program and the computer-readable medium through the processor and/or memory, and is used to execute the above-mentioned ramp exit road condition detection method when executing the computer-readable medium.

As can be seen from the above, the ramp exit road condition detection system, method and computer-readable medium of the present invention mainly use the road condition detection device to track the first image of the target vehicle captured by the camera device in the target area through object tracking technology to identify whether the target vehicle is driving illegally, and automatically detect whether the target vehicle is driving illegally, so as to further capture the second image to search for evidence of the target vehicle. Quickly and accurately provide evidence to law enforcement agencies for subsequent processing.

1: Ramp exit road condition detection system

10: The first photography installation

20:Second camera device

30: Road condition detection device

31: Image recognition module

311: Grooving line identification unit

312: Road shoulder identification unit

313: Lane identification unit

32: Violation evidence collection module

321: Illegal Evidence Collection Unit

322: Historical Access Unit

323: data storage unit

9: Target vehicle

9': other vehicles

A: target area

a1: shoulder

a2: ramp exit road

a3: Grooving line

a4: general road

D1: erection distance

D2: real distance

F ₀ ~F ₃ 、(X ₀ ,Y ₀ )~(X ₃ ,Y ₃ ): corner position

H: erection height

S51 to S511: Steps

T1: part one

T2: Part Two

Fig. 1 is a schematic diagram of the structure of the ramp exit road condition detection system of the present invention;

Fig. 2 is the schematic diagram of the ramp exit of the present invention;

Fig. 3 is the schematic diagram of the cascade matching of the present invention;

Fig. 4A is a schematic diagram of an actual lane picture of a general road;

Fig. 4B is a schematic diagram of a new lane picture after a general road change; and

5A and 5B are schematic flowcharts of the method for detecting road conditions at ramp exits according to the present invention.

The implementation of the present invention is described below through specific specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification.

It should be noted that the structures, proportions, and sizes shown in the drawings attached to this specification are only used to match the content disclosed in the specification for the understanding and reading of those who are familiar with this technology, and are not used to limit the conditions for the implementation of the present invention, so they have no technical significance. At the same time, terms such as "one", "first", "second", "upper" and "lower" quoted in this specification are only for the convenience of description and clarity, and are not used to limit the scope of the present invention. Changes or adjustments in their relative relationships are deemed to be within the scope of the present invention if there is no substantial change in the technical content.

FIG. 1 is a schematic diagram of the structure of the ramp exit road condition detection system 1 of the present invention. As shown in Figure 1, the ramp exit road condition detection system 1 includes: a first photographing device 10, a second photographing device 20, and a road condition detection device 30 having an image recognition module 31 and a violation evidence collection module 32, and the first photography device 10 and the second photography device 20 are connected to the road condition detection device 30 in communication, wherein the image recognition module 31 includes a grooved line recognition unit 311, a shoulder recognition unit 312 and a lane recognition unit 313, and a violation evidence collection The module 32 includes a violation evidence collection unit 321 , a history retrieval unit 322 and a data storage unit 323 .

Specifically, the first photographing device 10 and the second photographing device 20 are devices such as video cameras and monitors with video recording functions, but are not limited thereto; When in the electronic equipment of the calculation mechanism, each module and its unit in the road condition detection device 30 can be software, hardware or firmware; if it is hardware, it can be a processing unit, processor, computer or server with data processing and computing capabilities;

In this embodiment, as shown in FIG. 2 , the first photographing device 10 and the second photographing device 20 are fixed or movable respectively arranged at the corresponding two ends of the target area A at the exit of the ramp, so that the shooting range of the first photographing device 10 and the second photographing device 20 covers the target area A, wherein the target area A includes a shoulder a1, a ramp exit road a2, a grooved line a3 and a general road a4, and the first photographing device 10 is arranged on one or more target vehicles 9 leaving the ramp exit road a2 to photograph The first image of the front direction (or the rear direction) of the target vehicle 9, and the second photographing device 20 is arranged at the place where the target vehicle 9 enters the ramp exit road a2 to capture the second image of the target vehicle 9 in the rear direction (or the front direction). Furthermore, the first photographing device 10 and the second photographing device 20 send back the captured first image and the second image to the road condition detection device 30 in real time.

In one embodiment, the first photographing device 10 and the second photographing device 20 are respectively separated from the road surface of a road section by an installation height H, and the first photographing device 10 and the second photographing device 20 are respectively separated from the target area A by an erection distance D1, wherein the erection height H is greater than the height of the highest vehicle allowed to pass through the road section, and the erection distance D1 is greater than the erection height H, thereby avoiding that the images captured by the first photographing device 10 and the second photographing device 20 for the target vehicle 9 are easily shielded by the front or/and rear vehicles. The first photographing device 10 and the second photographing device 20 can have better viewing angles.

In one embodiment, the grooved line recognition unit 311 is used to identify whether the target vehicle 9 crosses the grooved line a3 in the target area A according to the first image; the road shoulder identification unit 312 is used to identify whether the target vehicle 9 is driving on the road shoulder a1 in the target area A during the detection time period according to the first image; Lane switch.

Specifically, the grooved line recognition unit 311, the road shoulder recognition unit 312, and the lane recognition unit 313 perform object tracking processing based on the first image and an object tracking technology to track the target vehicle 9 driving in the target area A. In one embodiment, the object tracking technology used by the grooved line recognition unit 311, the road shoulder recognition unit 312 and the lane recognition unit 313 is to use AI (Artificial Intelligence, artificial intelligence) deep learning YOLO (You Only Look Once) algorithm (such as YOLOv4) to perform object detection according to the first image, so as to detect the target vehicle 9 (that is, the object) in the first image.

In one embodiment, the grooved line recognition unit 311, the road shoulder recognition unit 312, and the lane recognition unit 313 perform vehicle trajectory tracking on the target vehicle 9 through the Deep SORT (Simple Online and Realtime Tracking with a Deep Association Metric) algorithm, thereby judging the vehicle trajectory of the target vehicle 9. Among them, the biggest feature of the Deep SORT algorithm is to add appearance information to use ReID (Person Re-identification, pedestrian re-identification ) appearance model to extract the appearance features of the target vehicle 9, thereby reducing the object ID (identification code) corresponding to the object (target vehicle 9) generated during object detection, thereby reducing the number of times of object ID switching to improve object tracking accuracy.

The following is the flow of the object tracking technology of the present invention:

(1) Use the YOLO algorithm to perform image detection on the first image to detect the target vehicle 9 (i.e. object) in the first image and the object coordinates of its center position for subsequent tracking of the target vehicle 9 in the first image.

(2) Use the cascade matching in the Deep SORT algorithm to predict the vehicle trajectory of the target vehicle 9 in the first image. As shown in FIG. 3 , the detailed content of the cascade matching is schematically shown. In the first part T1, the similarity between the currently detected object (target vehicle 9) and the previously detected object is judged by calculating the appearance model (ReID) and motion model (Mahalanobis distance). Conversely, if the difference between the currently detected object and the previously detected object is small, then it is judged that the currently detected object (target vehicle 9) is the same object as the previously detected object, and the cost matrix (Cost Matrix) is limited by the gate matrix (Gate Matrix).

Moreover, the second part T2 is the data association step of cascade matching, wherein cascade matching is performed through the Hungarian Algorithm, and the Hungarian algorithm uses the cost matrix to calculate the similarity between each frame of the first image and the target vehicle 9 (object) in the previous frame, thereby performing cascade matching to confirm whether each frame matches the target vehicle 9 in the previous frame, and then obtain the cascade matching result of each frame and the previous frame, and the cascade matching presets an error threshold (for example, max age is 7 0, that is, from missing age=0 to missing age=70), therefore, the target vehicle 9 (object) will have 70 recognition errors in the first image, and when it is confirmed that the cascade matching result of the target vehicle 9 in the two frames before and after the first image is unsuccessful, it means missing age=1, so as to compare each frame of the first image one by one with the previous frame. =70), the tracking target vehicle 9 will be abandoned. Conversely, when there is a successful match to the target vehicle 9 (that is, missing age<70), through By identifying the target vehicle 9 in the first image to predict its vehicle track, in another embodiment, the target vehicle 9 that has not lost the vehicle track is preferentially matched, and the blocked target vehicle 9 is retrieved by preferentially matching the target vehicle 9 that has not lost the vehicle track, thereby reducing the number of times of object ID switching due to the target vehicle 9 that is blocked and then reappears. For example, the matching of the target vehicle 9 in the second frame of the first image with the target vehicle 9 in the first frame of the previous frame is completed through the Hungarian algorithm, so as to obtain the cascade matching result of the first frame and the second frame.

(3) Use IOU (Intersection over Union) matching to compare the second frame in the first image with the first frame in the first image, that is, compare the picture of the second frame with the picture of the first frame of the previous frame to generate the overlap ratio between the picture of the second frame and the picture of the first frame. Specifically, the intersection of the Detection Result (that is, the picture of the second frame) and the Ground Truth (that is, the picture of the first frame) is compared with its union to calculate the overlap rate, which is the IOU matching result, and the formula of the IOU is as follows:

(4) The Kalman Filter is updated according to the cascade matching results of the first frame and the second frame, and the IOU matching results of the first frame and the second frame, and then the updated Kalman Filter results are used to calculate the cascade matching and IOU matching of other frames in the first image, so that the accuracy of the cascading matching results and IOU matching results can be increased. In other words, the Deep SORT algorithm can predict the vehicle trajectory through the updated Kalman Filter, thereby improving the accuracy of the vehicle trajectory.

In one embodiment, when the grooved line recognition unit 311 performs object tracking processing based on the first image and combined with the object tracking technology to identify that the target vehicle 9 crosses the grooved line a3 in the target area A, the road shoulder recognition unit 312 performs object tracking based on the first image and combined with the object tracking technology. Tracking processing to identify that the target vehicle 9 is driving on the shoulder a1 in the target area A during the detection time period, or/and the lane recognition unit 313 performs object tracking processing based on the first image and combined with object tracking technology to identify that the target vehicle 9 is not located on the off-ramp road a2 or/and general road a4 of the target area A. Evidence violations.

For example, when the grooved line identification unit 311 recognizes that the vehicle position of the target vehicle 9 enters the grooved line a3 from the ramp exit road a2 or/and the general road a4, it is determined that the target vehicle 9 has crossed the grooved line; The road shoulder; and the lane recognition unit 313 first judges whether the ramp exit road a2 or/and the general road a4 is a road section that can switch lanes. If the ramp exit road a2 or/and the general road a4 is a road section that cannot switch lanes, then when the lane recognition unit 313 recognizes that the target vehicle 9 leaves the original area (the ramp exit road a2 or the general road a4), it is determined that the target vehicle 9 is violating lane switching. On the contrary, if the ramp exit road a2 or/and the general road a4 is a road section that can switch lanes, However, the lane recognition unit 313 recognizes that the target vehicle 9 does not keep a safe distance from other vehicles 9', that is, it is determined that the target vehicle 9 is illegally switching lanes.

In another embodiment, when the lane recognition unit 313 recognizes that the target vehicle 9 is located in the ramp exit road a2 or/and the general road a4 of the target area A and is switching lanes, the lane recognition unit 313 further detects whether the target vehicle 9 maintains a safe driving distance with other vehicles (such as other vehicles 9' as shown in FIG. 4A ) when switching lanes.

Specifically, the lane identification unit 313 first performs perspective transformation (Perspective Transformation) on the actual lane picture of the ramp exit road a2 or/and the general road a4 in the target area A in the first image captured by the first photographing device 10. For example, the actual lane picture of the general road a4 shown in FIG.₀to F₁, F₁to F₂, F₂to F₃, F₃to F₀) of the actual distance for perspective transformation, perspective transformation is to project the picture to a new plane, also known as projection mapping, so as to transform the trapezoidal actual lane picture of the general road a4 into a rectangular new lane picture (as shown in Figure 4B), and the four corner positions of the actual lane picture are F₀~F₃Convert to the four corner positions of the new lane picture through the perspective transformation formula (X₀,Y₀)~(X₃,Y₃).

In detail, the perspective transformation formula adopted by the lane identification unit 313 is as follows:

,

。 Among them, u and v are the actual lane picture, corresponding to the coordinates (X, Y) of all objects in the transformed picture, and

,

.

可以拆成三個部分，

係為線性變換，例如：scaling(縮放)、shearing(剪切)或/及rotation(旋轉)、[F₃₁ F₃₂]係用於平移，[F₁₃ F₂₃] ^T 係產生透視變換，藉此改寫透視變換公式，改寫後之透視變換公式，如下所示： Furthermore, the transformation matrix

can be broken down into three parts,

It is a linear transformation, for example: scaling (zooming), shearing (cutting) or/and rotation (rotation), [F ₃₁ F ₃₂ ] is used for translation, [F ₁₃ F ₂₃ ] ^T is used to generate perspective transformation, so as to rewrite the perspective transformation formula. The rewritten perspective transformation formula is as follows:

Therefore, the lane identification unit 313 obtains the real positions of the target vehicle 9 and other vehicles 9' through the new lane picture converted from the general road a4 and combines the object coordinates of the target vehicle 9 obtained by the YOLO algorithm and the object coordinates of other vehicles 9', so that the lane identification unit 313 calculates the real distance D2 between the target vehicle 9 and other vehicles 9' based on the real positions of the target vehicle 9 and other vehicles 9' and using the Euclidean distance. The real distance D2 of ’ is less than the set safe driving distance (i.e. a preset distance), and when the image recognition technology is used to determine that the target vehicle 9 intends to switch to the same lane as other vehicles 9 ′, it can be determined that the target vehicle 9 is an event of switching lanes without maintaining a safe driving distance (i.e. a preset distance), so that the image recognition module 31 also produces a violation identification result to notify the violation evidence collection module 32 to collect violation evidence for the target vehicle 9.

In one embodiment, when the violation evidence collection module 32 receives the violation identification result from the image recognition module 31, the violation evidence collection module 32 instructs the second photographing device 20 to capture a second image in the direction of the rear of the target vehicle 9 for violation evidence collection, and then the second photographing device 20 uses the second image to perform license plate recognition and image capture functions to obtain the target vehicle 9's license plate, driving route, or/and violation time and other evidence of violation. In detail, the second photographing device 20 is based on the vehicle information (including vehicle appearance, vehicle color or license plate) of the target vehicle 9 in the first image captured by the first photographing device 10 to confirm the target vehicle 9 that needs to be collected for violation evidence, whereby the second photographing device 20 takes a second image of the target vehicle 9.

Furthermore, the violation evidence collection unit 321 integrates the second image of the target vehicle 9, the violation evidence, and the violation recognition result (including the first image) of the image recognition module 31, and forms the violation driving information of the target vehicle 9, which is displayed on a screen in real time for the management personnel to monitor the target vehicle 9. In addition, the illegal driving information of the target vehicle 9 is stored in the data storage unit 323 for management When personnel need to access data, the history access unit 322 provides management personnel to access the illegal driving information of the target vehicle 9 from the data storage unit 323, so that the record of illegal driving information can be used as a reference for law enforcement and accountability when an accident occurs afterwards, so as to protect the safety and road rights of passers-by.

FIG. 5A and FIG. 5B are schematic flow charts of the method for detecting road conditions at ramp exits according to the present invention, and are described with reference to FIG. 1 and FIG. 2 .

In step S51 , a first photographing device 10 is installed at a place where a target vehicle 9 exits the ramp exit road a2 to capture a first image of the target vehicle 9 in the direction of its head.

In step S52, the grooved line recognition unit 311 of the image recognition module 31 is used to perform object tracking processing based on the first image and in combination with object tracking technology to identify whether the target vehicle 9 crosses the grooved line a3 in the target area A. If it crosses the grooved line a3, then execute step S57; otherwise, if it does not cross the grooved line a3, continue to identify the target vehicle 9 or other vehicles.

In step S53, the road shoulder recognition unit 312 of the image recognition module 31 is used to judge whether it is in the detection time zone according to the first image, wherein, if it is in the detection time zone, then execute step S54; otherwise, if it is not in the detection time zone, continue to judge whether it is in the detection time zone.

In step S54, the road shoulder identification unit 312 is used to perform object tracking processing based on the first image and combined with object tracking technology to identify whether the target vehicle 9 is driving on the road shoulder a1 in the target area A, wherein, if driving on the road shoulder a1, then perform step S57; otherwise, if not driving on the road shoulder a1, continue to identify the target vehicle 9 or other vehicles.

In step S55, the lane recognition unit 313 of the image recognition module 31 is used to perform object tracking processing based on the first image and combined with object tracking technology to identify whether the target vehicle 9 is No Lane switching is performed on the switchable lane on the ramp exit road a2 or/and the general road a4 in the target area A, wherein, if the lane switch is performed on the switchable lane, then step S56 is performed; otherwise, if the lane switch is not performed on the switchable lane, then step S57 is performed.

In step S56, the lane identification unit 313 detects whether a safe driving distance is maintained between the target vehicle 9 and other vehicles. If there is a safe driving distance maintained, the lane identification unit 313 continues to identify the target vehicle 9 or other vehicles; otherwise, if the safe driving distance is not maintained, step S57 is performed.

In step S57 , the image recognition module 31 generates a violation recognition result to notify the violation evidence collection module 32 to collect violation evidence for the target vehicle 9 .

In step S58, when the violation evidence collection module 32 receives the violation recognition result from the image recognition module 31, it makes the second photographing device 20 take a second image in the direction of the rear of the target vehicle 9, so as to use the second image to perform license plate recognition and image capture functions to obtain the target vehicle 9's license plate, driving road or/and violation time and other evidence of violations.

In step S59, the violation evidence collection unit 321 of the violation evidence collection module 32 displays the violation driving information of the target vehicle 9 formed by the second image of the target vehicle 9, the violation evidence, and the violation identification result of the image recognition module 31 on a screen in real time.

In step S510 , the data storage unit 323 of the violation evidence collection module 32 stores the violation driving information of the target vehicle 9 .

In step S511 , the history retrieval unit 322 of the violation evidence collection module 32 provides monitoring personnel to retrieve the violation driving information of the target vehicle 9 from the data storage unit 323 .

In addition, the present invention also discloses a computer-readable medium, which is applied to a computing device or computer having a processor (eg, CPU, GPU, etc.) and/or memory, and stores instructions, And the computing device or computer can be used to execute the computer-readable medium through the processor and/or memory, so as to execute the above-mentioned method and each step when executing the computer-readable medium.

To sum up, the ramp exit road condition detection system and method of the present invention and its computer-readable medium use the grooved line recognition unit, the shoulder recognition unit and the lane recognition unit to track the first image of the target vehicle captured by the photography device in the target area to identify whether the target vehicle is driving illegally, and automatically detect whether the target vehicle is driving illegally, so as to further capture the second image to search for evidence of the target vehicle. driving conditions, and then quickly and accurately provide evidence to law enforcement agencies for subsequent processing of violating vehicles.

In addition, the ramp exit road condition detection system, method and computer-readable medium of the present invention have the following advantages or technical effects.

1. The present invention identifies illegal driving through the shoulders of the target area, ramp exit roads, trough lines, and general roads, and uses the target area that is completely photographed by the camera device installed on the front and rear sides of the target area to identify the illegal driving of the target vehicle.

2. The present invention can detect the distance of one or more target vehicles on the lane at the same time, bring 24-hour all-weather monitoring and provide definite evidence, which can be used as key evidence for law enforcement agencies to ban illegal driving and issue penalties.

3. The present invention can flexibly set the opening conditions of road shoulders, ramp exit roads or/and general roads according to the detection time zone and confirm whether to open lanes, and can adjust the detection time zone or switch lanes at any time according to law enforcement needs, which meets the needs of flexibility and rapid response required by technological law enforcement.

4. The present invention can accurately track the vehicle trajectory of the target vehicle through the YOLO algorithm combined with the Deep SORT algorithm, so as to improve the accuracy of target vehicle detection and identification efficiency It is more efficient, and it can avoid the occurrence of untraceable situations caused by foreign objects blocking the camera device to shoot the target vehicle.

The above-mentioned embodiments are only illustrative to illustrate the principles and effects of the present invention, and are not intended to limit the present invention. Anyone skilled in the art can modify and change the above-mentioned embodiments without departing from the spirit and scope of the present invention. Therefore, the scope of protection of the rights of the present invention should be listed in the scope of the patent application.

1: Ramp exit road condition detection system

10: The first photography installation

20:Second camera device

30: Road condition detection device

31: Image recognition module

311: Grooving line identification unit

312: Road shoulder identification unit

313: Lane identification unit

32: Violation evidence collection module

321: Illegal Evidence Collection Unit

322: Historical Access Unit

323: data storage unit

Claims (7)

A road condition detection system for a ramp exit, comprising: a first photographing device that photographs a target area at the ramp exit and a first image of at least one target vehicle in the target area; a road condition detection device that communicates with the first photographing device to receive the first image, uses the first image to perform object tracking processing, and performs vehicle tracking on the target vehicle driving in the target area in the first image, so that when it is confirmed that the target vehicle is driving illegally in the target area, a violation identification result is generated; and a second photographing device, The road condition detection device is connected by communication, so that when the road condition detection device generates the violation identification result, it takes a second image of the target area and the target vehicle in the target area, so as to obtain the violation evidence of the target vehicle based on the second image, and then transmits the violation evidence to the road condition detection device, so as to integrate the second image of the target vehicle, the violation evidence and the violation identification result into a violation driving information. The road condition detection device includes an image recognition module, and the image recognition module further includes: a grooved line recognition unit, which performs the object tracking process according to the first image to identify whether the target vehicle crosses the grooved line in the target area; a shoulder recognition unit, which performs the object tracking process according to the first image to identify whether the target vehicle is driving on the shoulder of the target area in the detection time zone; and a lane recognition unit, which performs the object tracking according to the first image processing to identify whether the target vehicle is located on a ramp exit road or a switchable lane on a general road in the target area for lane switching, Wherein, when the target vehicle crosses the grooved line in the target area, the target vehicle is driving on the shoulder of the road in the target area during the detection time period, or the target vehicle is not located in the off-ramp road of the target area or a switchable lane on a general road for lane switching, the image recognition module generates the violation identification result. The road condition detection system for ramp exits as described in claim 1, wherein when the lane recognition unit recognizes that the target vehicle is located on a switchable lane on a ramp exit road or a general road in the target area for lane switching, the lane recognition unit performs perspective transformation on the actual lane picture of the ramp exit road or general road in the target area in the first image to transform into a new lane picture, and then obtains the positions of the target vehicle and other vehicles, and the lane recognition unit calculates whether a preset distance is maintained between the target vehicle and the other vehicles, When the preset distance is not maintained, the image recognition module generates the violation recognition result. The ramp exit road condition detection system as described in Claim 1, wherein the road condition detection device includes a violation identification module, which is used to receive the violation identification result, so that the second photographing device can capture the second image of the target vehicle, so that the violation driving information can be displayed on a screen in real time. A road condition detection method for a ramp exit, comprising: taking a first image of a target area at the ramp exit and at least one target vehicle in the target area by a first photographing device; receiving the first image by a road condition detection device, using the first image to perform object tracking processing, and then performing vehicle tracking on the target vehicle driving in the target area in the first image, so as to generate a violation identification result when it is confirmed that the target vehicle is driving illegally in the target area; and When the road condition detection device generates the violation identification result, a second photographing device takes a second image of the target area and the target vehicle in the target area to obtain the violation evidence of the target vehicle based on the second image, and then transmits the violation evidence to the road condition detection device to integrate the second image of the target vehicle, the violation evidence and the violation identification result into a violation driving information, wherein the target area is a shoulder, a ramp exit road, a trough line or a general road , and the road condition detection device includes an image recognition module having a grooved line recognition unit, a road shoulder recognition unit and a lane recognition unit, wherein, the grooved line recognition unit performs the object tracking process based on the first image to identify whether the target vehicle crosses the grooved line in the target area, and the road shoulder recognition unit performs the object tracking process based on the first image to identify whether the target vehicle is driving on the road shoulder in the target area during the detection time period, and the lane recognition unit performs the object tracking process based on the first image, and the lane recognition unit according to the The object tracking process is performed on the first image to identify whether the target vehicle is on an off-ramp road or a switchable lane on a general road in the target area for lane switching, wherein the image recognition module generates the violation identification result when the target vehicle crosses a grooved line in the target area, the target vehicle travels on a road shoulder in the target area during the detection time period, or the target vehicle is not located on an off-ramp road or a switchable lane on a general road in the target area. The road condition detection method for ramp exits as described in claim 4 further includes when the lane recognition unit recognizes that the target vehicle is located in the switchable lane of the ramp exit road or general road in the target area for lane switching, the lane recognition unit performs perspective transformation on the actual lane picture of the ramp exit road or general road in the target area in the first image to transform into a new lane picture, and then obtains the positions of the target vehicle and other vehicles, and the lane recognition unit calculates the distance between the target vehicle and the road. Whether to keep a preset distance between other vehicles, so that when the preset distance is not kept, the image recognition module generates the violation recognition result. The road condition detection method at the ramp exit as described in Claim 4, wherein the road condition detection device includes a violation identification module, so that when the violation identification module receives the violation identification result, the second photographing device is made to capture the second image of the target vehicle, so that the violation driving information can be displayed on a screen in real time. A computer-readable medium, used in a computing device or a computer, is stored with instructions to execute the method for detecting road conditions at ramp exits as described in any one of claims 4 to 6.

Images