TWI743729B - Traffic violation detection system and control method thereof - Google Patents

Abstract

A traffic violation detection system includes a radar device, a video recording device and a processing unit. The radar device generates radar sensing results. The video recording device generates the shooting image. The processing unit extracts the radar sensing result of the detected vehicle, calculates the radar sensing position data indicating the position of the vehicle based on the aforementioned radar sensing result, and converts each radar sensing position data into the coordinates of the captured image Department of expression. The processing unit extracts the captured image of the vehicle from the captured image according to the sensing time of the target radar sensing result and the predetermined compensation time. For the aforementioned captured images, the processing unit uses image segmentation technology in conjunction with radar to sense location data to generate vehicle location data. The processing unit determines whether the vehicle location data meets one of the multiple violation conditions.

Description

Traffic violation detection system and control method thereof

The invention relates to a detection system, in particular to a traffic violation detection system. The invention also relates to a control method of a traffic violation detection system.

The existing method of traffic violation detection is to use image processing to determine whether the vehicle violates the traffic rules, and after judging and confirming the violation of the traffic rules such as speeding, double-yellow line turning, reverse driving and other violations, the image of the offending vehicle is captured and the license plate is identified. Sending a ticket, but in rainy or night, the use of image processing will greatly reduce the accuracy and result in inefficient law enforcement. Therefore, how to improve the shortcomings of the prior art is a subject to be further explored in the present invention.

Therefore, the purpose of the present invention is to provide a traffic violation detection system to solve the problem of the decrease in the identification accuracy of the traffic violation system.

Another object of the present invention is to provide a control method of a traffic violation detection system.

Therefore, the traffic violation detection system of the present invention includes a radar device, a video recording device, and a processing unit in some embodiments.

The radar device is used to continuously sense a section to generate continuous multiple radars The sensing results, the radar sensing results have a first coordinate system.

The video recording device is used for continuously shooting the road section to generate a plurality of continuous shooting images, and the shooting images have a second coordinate system.

The processing unit is electrically connected to the radar device and the recording device, and the processing unit determines whether a vehicle is passing the road section according to the radar sensing results.

When the processing unit determines that the vehicle is traveling on the road section according to the radar sensing results, the processing unit extracts from the radar sensing results a plurality of target radar sensing results that detect the vehicle. Radar sensing results.

For each target radar sensing result, the processing unit calculates a radar sensing position data that indicates the position of the vehicle and is expressed in the first coordinate system, and converts each radar sensing position data to the second Coordinate system expression.

The processing unit extracts a plurality of shooting images of the vehicle as a plurality of target shooting images from the shooting images according to the sensing time of the target radar sensing results and a predetermined compensation time.

The processing unit shoots images for each target, uses image segmentation technology and cooperates with the radar sensing position data to generate a vehicle position data.

The processing unit determines whether the vehicle location data meets one of multiple violation conditions.

When the processing unit determines that the vehicle location data meets one of the violation conditions as the violation condition of the target violation condition, the processing unit is based on the target violation condition The captured image identifies a license plate number of the vehicle and generates a violation case data. The violation case data includes the license plate number, the target shooting images, a traffic law related to the violation condition, and an indication of the target shooting The violation time of the shooting time of the image.

In some implementations, it further includes a traffic signal device electrically connected to the processing unit. The traffic signal device generates a light signal history data. The light signal history data includes a plurality of light signal records. The record includes an output light signal content and an output time. The processing unit also refers to the light signal history data when judging whether the vehicle location data meets some of the violation conditions.

In some implementation aspects, the processing unit generates a vehicle type data indicating the vehicle type of the vehicle based on the target radar sensing results, wherein the processing unit is determining whether the vehicle location data meets some of the violation conditions When, also refer to the vehicle type information.

In some implementations, the processing unit generates vehicle type data indicating the vehicle type of the vehicle based on the target captured images, wherein the processing unit determines whether the vehicle location data meets some of the violation conditions, Also refer to the vehicle type information.

In some implementation aspects, the processing unit generates vehicle type data indicating the vehicle type of the vehicle based on the images taken by the targets, the radar sensing results of the targets, and an image correction weight, wherein the processing unit is determining the vehicles Is the location data When some of these violation conditions are met, the vehicle type information is also referred to.

The control method of the traffic violation detection method of the present invention is implemented by a traffic violation detection system. The traffic violation detection system includes a radar device, a video recording device, and a processing unit. The radar device is used to continuously sense a section to generate continuous The video recording device is used to continuously shoot the road section to generate continuous shooting images, the shooting images have a second coordinate system, and the processing unit is electrically connected to the radar device and the video recording device , The method includes: the processing unit determines whether a vehicle is passing the road section according to the radar sensing results; when the processing unit determines that the vehicle is passing the road section according to the radar sensing results, the processing unit senses from the radars From the detection result, a plurality of radar sensing results of the vehicle detected as a plurality of target radar sensing results are extracted; for each target radar sensing result, the processing unit calculates the position of the vehicle and uses the target radar sensing result. A radar sensing position data expressed by the first coordinate system, and each radar sensing position data is converted into the second coordinate system; the processing unit is based on the sensing time of the target radar sensing results and a predetermined Compensation time, to extract multiple shot images of the vehicle as multiple target shot images from the shot images; the processing unit shoots images for each target, uses image segmentation technology and matches the radar sensing Location data to generate a vehicle location data; the processing unit determines whether the vehicle location data meets one of the multiple violation conditions; when the processing unit determines that the vehicle location data meets one of the violation conditions as the target violation condition Violation of the conditions, the processing unit shoots according to these targets The image identifies a license plate number of the vehicle, and generates a violation case data, the violation case data includes the license plate number, the target shot images, a traffic law related to the violation condition, and an indication of the target shot images The offending time of the shooting time.

The present invention has at least the following effects: the processing unit converts the radar sensing position data expressed in the first coordinate system into the second coordinate system, and sensing based on the target radar sensing results The time and the predetermined compensation time are taken from the shooting images to capture the target shooting images of the vehicle, and the image segmentation technology is used in conjunction with the radar sensing position data to generate the shooting images used to determine whether the violation conditions are met The vehicle location data enables the present invention not only to rely on the processing unit to perform image processing on the captured images generated by the recording device to determine violation conditions, but also to combine the radar sensing results to improve rain or night alone. The problem of poor image recognition accuracy. In addition, the present invention also refers to the light signal history data and the vehicle type data when judging whether the vehicle location data meets some of the violation conditions by the processing unit. It can not only judge the violations related to the driving path of the vehicle, such as double yellow line turning or reverse driving, but also judge by the historical data of the light signal and the data of the vehicle type, such as running a red light or turning right at a red light, etc. and signs Related violations and, for example, violations related to the vehicle type without a two-stage left turn.

1: radar device

2: Recording device

3: Processing unit

4: Traffic signal equipment

100: Traffic Violation Detection System

S01~S09: steps

The other features and effects of the present invention will be described in the embodiment with reference to the drawings It is clearly presented in Fig. 1 is a schematic diagram of a hardware connection relationship of an embodiment of the traffic violation detection system of the present invention; and Figs. 2 to 3 are a flowchart of this embodiment.

Before the present invention is described in detail, it should be noted that in the following description, similar elements are denoted by the same numbers.

Referring to FIG. 1, an embodiment of a traffic violation detection system 100 of the present invention includes a radar device 1, a video recording device 2, a processing unit 3, and a traffic sign device 4.

The radar device 1 is used for continuously sensing a section to generate a plurality of continuous radar sensing results, and the radar sensing results have a first coordinate system. The radar sensing results can display the distance between the detected object and the radar device 1 in reality through the first coordinate system. In this embodiment, the radar device 1 is a millimeter wave radar with an operating frequency of 79 gigahertz (GHz), but it is not limited to this.

The video recording device 2 is used for continuously shooting the road section to generate a plurality of continuous shooting images, and the shooting images have a second coordinate system.

The traffic signal device 4 is used to generate a light signal history data. The light signal history data includes a plurality of light signal records, and each light signal record includes an output light signal content and an output time. The content of the output light can be, for example, red light, yellow light, green light, left turn special Use lights and so on.

The processing unit 3 is electrically connected to the radar device 1, the recording device 2 and the traffic signal device 4 to receive information generated by the radar device 1, the recording device 2 and the traffic signal device 4.

1 to 3, the following describes the steps performed by the traffic violation detection system 100 in this embodiment. First, as shown in step S01, the processing unit 3 receives the radar sensing results generated by the radar device 1, the captured images generated by the video recording device 2, and the light signals generated by the traffic signal device 4. Record.

Then, as shown in step S02, the processing unit 3 determines whether a vehicle is passing the road section based on the radar sensing results. If the processing unit 3 determines that a vehicle is passing the road section, then step S03 is executed. If the processing unit 3 If it is judged that there is no vehicle passing the road section, the process ends. For example, if the processing unit 3 determines that the radar sensing results match one of a plurality of pre-stored vehicle radar patterns indicating vehicles of different vehicle types in a radar sensing database, it determines that the vehicle is traveling on the road section.

As shown in step S03, when the processing unit 3 determines that the vehicle is traveling on the road section based on the radar sensing results, the processing unit 3 extracts from the radar sensing results that the vehicle is sensed as Radar sensing results of multiple target radar sensing results.

Then, as shown in step S04, the processing unit 3 for each target radar sensing result calculates a mine that indicates the position of the vehicle and is expressed in the first coordinate system. To reach the sensed position data, and each radar sensed position data is first converted into a Real World Coordinate System (Real World Coordinate System), and then converted into the second coordinate system.

Then, as shown in step S05, the processing unit 3 captures the multiple targets captured by the vehicle from the captured images according to the sensing time of the target radar sensing results and a predetermined compensation time. For the captured image of the captured image, one of the sensing time of the target radar sensing results can be, for example, "2020/03/01 14:05:35", etc., but it is not limited by the foregoing disclosure method. Specifically, because the millimeter wave radar calculates the distance by reflection time, the result it knows will be affected by the distance between the object and the radar and the calculation time. The data obtained is not the data of the current time, so the processing If the shooting time of the images captured by the unit 3 from the shooting images is to be the same as the sensing time of the target radar sensing results, the time difference must be compensated through the predetermined compensation time.

Then, as shown in step S06, the processing unit 3 shoots an image for each target, uses image segmentation technology and cooperates with the radar-sensed position data to generate a vehicle position data. Specifically, image segmentation technology uses deep learning to increase the accuracy of image recognition, and also uses image processing technology to find out the region of interest in the image space. Reduce processing time. By combining the radar sensing location data, the present invention is not only used for image processing detection than the existing traffic violation detection system. The vehicle location will also be combined with radar sensing data to further improve the accuracy of the present invention for judging the vehicle location during rainy days or at night.

Then, as shown in step S07, the processing unit 3 generates vehicle type data indicating the vehicle type of the vehicle based on the target radar sensing results. In this embodiment, the processing unit 3 generates vehicle type data based on the target radar sensing results The method of the vehicle type data is to use deep learning to perform point group processing to calculate the probability value of each vehicle being a variety of vehicle types and to judge the vehicle type based on this. For example, if the vehicle knows that the probability of being a locomotive is 0.8 and the probability of being a car is 0.2 through point group processing, then it is determined that the vehicle is a locomotive, but the foregoing example is not limited to this.

In other embodiments, the processing unit 3 generates a vehicle type data indicating the vehicle type of the vehicle according to the target shot images.

In other embodiments, the processing unit 3 generates a vehicle type data indicating the vehicle type of the vehicle based on the target radar sensing results, the target captured images, and an image correction weight. Specifically, the processing unit 3 first multiplies the vehicle type probabilities calculated based on the target images taken by the image correction weight, and respectively adds the vehicle type probabilities calculated based on the target radar sensing results. According to the aforementioned parameters, the vehicle type is judged to generate the vehicle type data. For example, if the processing unit 3 determines that the vehicle is a locomotive with a probability of 0.7 and a car is 0.3 based on the target radar sensing results; the probability of determining that the vehicle is a locomotive based on the target images is 0.4, The probability of being a car is 0.6; the image correction weight is set to target shooting at night The image judges that the weight of the locomotive is 0.8 and the weight of the car is 0.7; the processing unit 3 calculates that the parameter of the vehicle is a locomotive is 0.7+0.4*0.8=1.02, and the parameter of the car is 0.3+0.6*0.7=0.72, the The processing unit 3 generates the vehicle type data of the locomotive according to the above result, but not limited to the foregoing example, wherein the image correction weight will be changed according to time or weather.

Then, as shown in step S08, the processing unit 3 refers to the light signal history data and the vehicle type data to determine whether the vehicle location data meets one of the multiple violation conditions, and if the vehicle location data meets one of the multiple violation conditions For one, step S09 is then executed. If the vehicle location data does not meet one of the multiple violation conditions, the process ends. For example, by using the vehicle location data obtained in step S06, the vehicle's traveling trajectory is formed, and it is determined whether the traveling trajectory meets the violation trajectory related to one of the violation conditions, thereby, when the The processing unit 3 can distinguish when the vehicle is driving in the reverse direction, turning in a double yellow line, turning in violation of the rules, turning left without two stages, turning over the line, turning right at a red light, or running through a red light.

As shown in step S09, when the processing unit 3 determines that the vehicle location data meets one of the violation conditions as the violation condition of the target violation condition, the processing unit 3 recognizes a license plate number of the vehicle according to the images taken by the target , And generate a violation case data, the violation case data includes the license plate number, the target shooting images, a traffic law related to the violation conditions, and a violation time indicating the shooting time of the target shooting images, and Upload the violation case data to the traffic center for Complete the action of the post.

In summary, the effect of the present invention is that the processing unit 3 converts the radar-sensing position data expressed in the first coordinate system into the second coordinate system, and performs the radar sensing according to the target The resulting sensing time and the predetermined compensation time extract the target shot images of the vehicle from the shot images, and use the image segmentation technology and the radar sensing position data to generate for judging compliance with regulations Conditions of the vehicle position data, the present invention not only depends on the processing unit 3 to perform image processing on the captured images generated by the recording device 2 to determine the violation conditions, but also combines the radar sensing results to To improve the problem of poor accuracy of image recognition in rainy days or nights alone, in addition, the present invention uses the processing unit 3 to refer to the light signal history data and when determining whether the vehicle location data meets some of the violation conditions. The vehicle type data enables the present invention not only to determine violations related to the driving path of the vehicle, such as double-yellow line turning or reverse driving, but also to determine, for example, running a red light or a red light based on the historical data of the light signal and the vehicle type data. Turning right and other violations related to signs and violations related to vehicle types such as turning left without a two-stage turn can indeed achieve the purpose of the invention.

However, the above are only examples of the present invention. When the scope of implementation of the present invention cannot be limited by this, all simple equivalent changes and modifications made in accordance with the scope of the patent application of the present invention and the content of the patent specification still belong to Within the scope covered by the patent of the present invention.

1········ Radar Device 2········ Video Recording Device 3········Processing unit 4········ Traffic signal equipment 100····· Traffic Violation Detection System

Claims (8)

A traffic violation detection system includes: a radar device for continuously sensing a section to generate a plurality of continuous radar sensing results, the radar sensing results having a first coordinate system; a video recording device for continuously The road section is photographed to generate consecutive multiple photographed images, the photographed images have a second coordinate system; a traffic signal device is used to generate a light signal history data, the light signal history data includes multiple light signal records, Each light signal record includes an output light signal content and an output time; and a processing unit, which is electrically connected to the radar device, the recording device, and the traffic signal equipment; the processing unit judges based on the radar sensing results Whether a vehicle is traveling on the road section; when the processing unit determines that the vehicle is traveling on the road section based on the radar sensing results, the processing unit extracts the plurality of radar sensing results that detect the vehicle as multiple Radar sensing results of a target radar sensing result; for each target radar sensing result, the processing unit calculates a radar sensing position data that indicates the position of the vehicle and expressed in the first coordinate system, and calculates each target radar sensing result. A radar sensing position data is converted into the second coordinate system; the processing unit extracts the captured images from the captured images according to the sensing time of the target radar sensing results and a predetermined compensation time Multiple images of the vehicle are taken as images taken by multiple targets; the processing unit takes images for each target, uses image segmentation technology and cooperates with the radar sensing position data to generate a vehicle position data; the processing unit uses The vehicle position data forms the vehicle's traveling trajectory; the processing unit determines whether the traveling trajectory meets the illegal trajectory of one of the multiple violation conditions, wherein, when determining, the processing unit also refers to the light signal history data; when The processing unit determines that the travel trajectory meets the violation trajectory of one of the violation conditions as the target violation condition. The processing unit recognizes a license plate number of the vehicle according to the images taken by the targets, and generates a violation case data, The violation case data includes the license plate number, the images captured by the targets, a traffic law related to the violation conditions, and a violation time indicating the shooting time of the images captured by the targets. The traffic violation detection system according to claim 1, wherein the processing unit generates a vehicle type data indicating the vehicle type of the vehicle according to the target radar sensing results; wherein the processing unit is determining whether the vehicle location data conforms to For some of these violations, reference is also made to the vehicle type information. The traffic violation detection system according to claim 1, wherein the processing unit generates a vehicle type indicating the vehicle type of the vehicle according to the images taken by the targets Data; where the processing unit also refers to the vehicle type data when determining whether the vehicle location data meets some of the violation conditions. The traffic violation detection system according to claim 1, wherein the processing unit generates a vehicle type data indicating the vehicle type of the vehicle based on the images taken by the targets, the radar sensing results of the targets, and an image correction weight; wherein, the The processing unit also refers to the vehicle type data when determining whether the vehicle location data meets some of the violation conditions. A traffic violation detection method is implemented by a traffic violation detection system. The traffic violation detection system includes a radar device, a video recording device, a traffic sign device, and a processing unit. The radar device is used to continuously sense Generate a plurality of continuous radar sensing results, the radar sensing results have a first coordinate system, the recording device is used to continuously shoot the road section to generate a plurality of continuous shooting images, the shooting images have a second coordinate The traffic signal equipment is used to generate a light signal history data, the light signal history data includes a plurality of light signal records, each light signal record includes an output light signal content and an output time, the processing unit is electrically connected to The radar device, the video recording device, and the traffic signal equipment, the method includes: the processing unit determines whether a vehicle is passing the road section according to the radar sensing results; when the processing unit determines according to the radar sensing results The vehicle travels on the road section, and the processing unit extracts from the radar sensing results that the vehicle is sensed as multiple target radar sensing results. Radar sensing results; for each target radar sensing result, the processing unit calculates a radar sensing position data that indicates the position of the vehicle and expressed in the first coordinate system, and converts each radar sensing position data The result is expressed in the second coordinate system; the processing unit extracts from the captured images the plurality of captured images of the vehicle as a plurality of targets based on the sensing time of the target radar sensing results and a predetermined compensation time The captured image of the captured image; the processing unit captures the image for each target, uses image segmentation technology and matches the radar sensing position data to generate a vehicle position data; the processing unit forms the vehicle position data by using the vehicle position data Traveling trajectory; the processing unit determines whether the traveling trajectory meets the violation trajectory of one of the multiple violation conditions, wherein, when determining, the processing unit also refers to the historical data of the light signal; when the processing unit determines that the traveling trajectory meets the If one of the violation conditions is regarded as the violation trajectory of the violation condition of the target violation condition, the processing unit recognizes a license plate number of the vehicle according to the images taken by the targets, and generates a violation case data. The violation case data includes the license plate number, The target shooting images, a traffic law related to the violation condition, and a violation time indicating the shooting time of the target shooting images. The method for detecting traffic violations according to claim 5, wherein the processing unit According to the target radar sensing results, a vehicle type data indicating the vehicle type of the vehicle is generated; wherein the processing unit also refers to the vehicle type data when determining whether the vehicle location data meets some of the violation conditions. According to the traffic violation detection method of claim 5, the processing unit generates a vehicle type data indicating the vehicle type of the vehicle according to the images taken by the targets; wherein the processing unit is determining whether the vehicle location data conforms to the vehicle type data. In case of violation of some of the conditions, the vehicle type information is also referred to. The traffic violation detection system according to claim 5, wherein the processing unit generates a vehicle type data indicating the vehicle type of the vehicle based on the images taken by the targets, the radar sensing results of the targets, and an image correction weight; wherein, the The processing unit also refers to the vehicle type data when determining whether the vehicle location data meets some of the violation conditions.

Images