A method and system for determining coincidence between vehicle body points and scene points

Abstract

The present invention relates to the field of driving teaching technology, and more specifically, to a method and system for determining the overlap between vehicle body points and scene points. The method includes recording a driving perspective video with a head-mounted camera, configuring vehicle body points and storing them in a database, analyzing scene points through the driving perspective recorded by the head-mounted camera, inputting the scene points into the database, calculating the overlap between scene points and vehicle body points, outputting the overlap between each point and reporting it, and transmitting the driving perspective video with point marks to the instructor through an external display module. The driving perspective recorded by the present invention not only enables a voice broadcast reminder when the driving trainee crosses the line, thus preventing the driving trainee from not being able to find and correct their own problems in time, but also allows the instructor to determine whether the current trainee has crossed the line through the driving perspective, so that the trainee can quickly understand the defects in driving, improve the pertinence of subsequent teaching, and contribute to the quality of teaching.

Description

Technical Field

[0001] The present invention relates to the technical field of driving teaching, and in particular to a method and system for determining the coincidence of vehicle body points and scene points. Background Art

[0002] With the development of technology, cars have gradually entered ordinary families. When driving a car, users need to obtain a driver's license in order to legally drive on the road. Novice drivers often enroll in driving schools during training. Driving schools are training institutions that teach drivers how to drive a vehicle and driving ethics, and assist them in passing the examination of the vehicle management department to obtain a driver's license. When driving school instructors train students, they usually ask students to memorize the body points and scene points. By checking whether the body points and scene points coincide, students can judge whether they have crossed the line when driving in a straight line. This is also the basis of learning to drive a car. However, novice drivers have poor driving sense and are not proficient in memorizing the points when trying to drive for the first time. As a result, it is not convenient for novice drivers to detect when they have crossed the line in time, which affects the efficiency of driving learning. In addition, it is not convenient for the instructor to observe the points from the student's driving perspective while sitting in the passenger seat, resulting in an inability to accurately determine which point the student has misremembered, which affects the teaching progress. In view of this, we propose a method and system for determining the coincidence of body points and scene points. Summary of the Invention

[0003] The object of the present invention is to provide a method and system for determining the coincidence of vehicle body points and scene points, so as to solve the problems raised in the above background technology.

[0004] To solve the above technical problems, one of the objectives of the present invention is to provide a method for determining whether a vehicle body point and a scene point coincide with each other, comprising the following steps:

[0005] S1. Wear a head-mounted camera on the driver's head to record the driving perspective video;

[0006] S2. Embed a point database in the head-mounted camera and store the vehicle body points in the database;

[0007] S3. Analyze scene points using the driving perspective recorded by the head-mounted camera, and mark the scene points and vehicle body points in the database in the driving perspective video;

[0008] S4. Input the scene points into the database and calculate the overlap between the scene points and the vehicle body points;

[0009] S5. Output the overlapping status of each point and broadcast it. At the same time, transmit the driving perspective video with point marks to the instructor through the external display module.

[0010] Preferably, the head-mounted camera in S1 is fixed on the driver's head using a strap, and the camera is a motion camera.

[0011] Preferably, the database in S2 adopts a chain memory, and its expression is:

[0012] A=

(a1,A1),(a2,A2),...,(ak,Ak)

[0013] Where A represents the set of vehicle body points configured in the database, a1, a2, ..., ak represent the positions of the vehicle body points, A1, A2, ..., Ak represent the state after a1 overlaps with the scene points, and k is the number of points.

[0014] Preferably, the analysis of scene points in S3 adopts a lane line recognition algorithm, including the following steps:

[0015] S3.1. After selecting a frame of road image from the driving perspective video, obtain a road grayscale image;

[0016] S3.2. Perform image segmentation on the grayscale image: Select a grayscale value and use it as the boundary to divide the image pixel grayscale values ​​into two categories, large and small. Calculate the number of pixels in the two categories and their average values. Then calculate their inter-class variance. The grayscale value corresponding to the maximum inter-class variance is taken as the segmentation threshold.

[0017] S3.3. Use structural elements to detect the segmented image, use morphological gradients for edge detection, filter out some noise points, and detect two parallel straight lines as lane lines.

[0018] Preferably, the calculation formula of S3.2 is:

[0019] w(i)=n1(i)n2(i)[v1(i)-v2(i)] 2

[0020] thre=ArgMax(w(i))

[0021] Among them, w(i) is the grayscale value of the image pixel, n1(i) is the number of pixels with a grayscale value less than i, n2(i) is the number of pixels with a grayscale value greater than or equal to i, v1(i) and v2(i) are the average values ​​of the number of pixels, respectively, and thre is the subscript corresponding to when w(i) takes the maximum value, that is, the segmentation threshold. After obtaining the road grayscale image, in order to separate the target (lane line) and the background (non-lane line), image segmentation processing is required. The key to image segmentation is to select an appropriate threshold. Improper threshold selection will cause image details to be blurred or lost, etc. Therefore, it is more accurate to determine the segmentation threshold through the above calculation formula.

[0022] Preferably, the calculation of the coincidence of scene points and vehicle body points in S4 adopts an image coordinate comparison algorithm, including the following steps:

[0023] Call up a frame from the driving perspective video, divide the image into multiple equal regions, establish coordinate axes based on the image size, and define the coordinate point set of the vehicle body point line as: (X, Y) = [(X1, Y1), (X2, Y2), ...], and the coordinate point set of the scene point as: (x, y) = [(x1, y1), (x2, y2), ...], if (X, Y) ≈ (x, y), it represents the scene point and vehicle body point.

[0024] A second object of the present invention is to provide a vehicle body point and scene point coincidence determination system, including any one of the vehicle body point and scene point coincidence determination methods described above, including a database, a video acquisition unit, a scene point analysis unit, a coincidence calculation unit and a signal output unit, wherein the database output end is respectively connected to the video acquisition unit and the coincidence calculation unit output end, the video acquisition unit output end is connected to the scene point analysis unit and the signal output unit output end, the scene point analysis unit output end is connected to the coincidence calculation unit output end, and the coincidence calculation unit output end is connected to the signal output unit output end, wherein:

[0025] The database is used to store the configured vehicle body points and the driving angle data of the camera acquisition unit;

[0026] The camera acquisition unit is used to wear a head-mounted camera to record driving perspective video;

[0027] The scene point analysis unit is used to analyze scene points through the driving perspective video stored in the database, and mark the scene points and vehicle body points in the database in the driving perspective video;

[0028] The overlap calculation unit is used to calculate the overlap between the scene points of the scene point analysis unit and the vehicle body points stored in the database;

[0029] The signal output unit is used to input the overlap status into the database, output the safety distance data for broadcast, and transmit the driving perspective video with point marks to the instructor through the external display module.

[0030] Preferably, the video acquisition unit includes a head-mounted camera and a driving perspective recording module, the head-mounted camera is used to shoot driving videos from a driving perspective, and the driving perspective recording module is used to temporarily store driving videos.

[0031] Preferably, the output end of the signal output unit is also connected to an external display module and a voice broadcast module. The external display module is used to transmit the driving perspective video to a display screen for display, so that the instructor can observe the driving perspective to teach the driving student. The voice broadcast module is used to give a voice broadcast reminder when the driving student crosses the line.

[0032] Compared with the prior art, the present invention has the following beneficial effects: a head-mounted camera is worn on the head of a driving trainee, body points are configured and stored in a database, the driving perspective is calculated to analyze the overlap between scene points and body points, the overlap between each point is output and broadcasted, and the driving perspective video with point marks is transmitted to the instructor through an external display module. The driving perspective recorded by the head-mounted camera not only enables the driving trainee to receive voice reminders when crossing the line, thus preventing the driving trainee from not being able to find and correct their own problems in time, but also cultivates the trainee's sense of safety in the car, and the instructor can judge whether the current trainee has crossed the line through the driving perspective, so that the trainee can quickly understand the driving defects, improve the targetedness of subsequent teaching, and benefit the teaching quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 This is a flowchart of the overall process of Example 1;

[0034] Figure 2 This is the overall structural principle diagram of Example 1;

[0035] Figure 3 A schematic diagram of configuring a vehicle body marking in the database of Example 1;

[0036] Figure 4 This is a demonstration diagram of the overlap calculation of Example 1;

[0037] Figure 5 This is a flow chart of the lane line recognition algorithm of Example 1;

[0038] The meaning of each number in the figure is:

[0039] 100. Database;

[0040] 200. Video acquisition unit;

[0041] 300, scene point analysis unit;

[0042] 400, coincidence calculation unit;

[0043] 500, signal output unit;

[0044] 510, external display module; 520, voice broadcast module. DETAILED DESCRIPTION

[0045] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0046] Example 1

[0047] like Figure 1-5 As shown, one of the purposes of the present invention is that with the development of science and technology, cars have gradually entered ordinary families. When driving a car, users need to obtain a driver's license so that they can legally drive on the road. Novice drivers often enroll in driving schools during training. Driving schools are training institutions that teach drivers how to drive a vehicle and driving ethics and assist them in passing the examination of the vehicle management department to obtain a driver's license. When driving school coaches train students, they usually ask students to remember the vehicle body points and scene points. By checking whether the vehicle body points and scene points coincide with each other, students can judge whether they have crossed the line when driving in a straight line. This is also the basis of learning to drive a car. However, novice drivers have poor car sense and are not proficient in memorizing points when trying to drive for the first time. As a result, it is inconvenient for novice drivers to find out in time when they have crossed the line, which affects the efficiency of driving learning. In addition, it is inconvenient for the coach to observe the points from the student's driving perspective when sitting in the passenger seat, resulting in an inability to accurately judge which point the student has remembered incorrectly, affecting the teaching progress. In view of this, a method for determining the coincidence of vehicle body points and scene points is provided, which includes the following steps:

[0048] S1. A head-mounted camera is placed on the driver's head to record the driver's perspective video. Since the driver needs to look forward when driving, the head-mounted camera faces forward to record the driver's perspective video. This allows the driver to record the driving video during the driving training process, making it easier to review and identify errors later.

[0049] The head-mounted camera in S1 is fixed on the driver's head with a strap. The camera is a motion camera. The working principle of the motion camera is known to those skilled in the art: light emitted to an object is reflected by the surface of the object, and the reflected light is transmitted to the image sensor through the lens. The image sensor receives the reflected light and converts the optical signal into an electrical signal and transmits it to the analog-to-digital conversion circuit; the analog-to-digital conversion circuit converts the received analog electrical signal into a digital electrical signal and transmits it to the digital signal processing chip for processing. The final processed signal is transmitted to the computer through the USB interface, and the original image can be displayed on the monitor.

[0050] S2. A point database is embedded in the head-mounted camera, and the vehicle body points are stored in the database. The vehicle body points previously taught by the driving school instructor are stored in the database. The vehicle body points are positions that are helpful for learning to drive safely based on previous teaching experience. By observing the overlap between the points on the vehicle body and the scene points, the driver can learn to drive faster.

[0051] Furthermore, the database in S2 uses chain storage, which is expressed as follows:

[0052] A=

(a1,A1),(a2,A2),...,(ak,Ak)

[0053] Where A represents the set of vehicle body points configured in the database, a1, a2, ..., ak represent the positions of the vehicle body points, A1, A2, ..., Ak represent the state after a1 overlaps with the scene points, and k is the number of points;

[0054] Specific as Figure 3 As shown: a1 is the position of the left side of the steering wheel, and A1 is the output of the left wheel pressure line of the car when a1 coincides with the lane line angle of the scene point;

[0055] a2 is the position on the right side of the steering wheel, and A1 is the state where a2 and the scene point lane line angle coincide with each other, and the output lane line is in the middle of the vehicle body;

[0056] a3 is the middle point of the car's center console, and A1 is the output of the car's right wheel crossing the line when a3 and the scene point lane line angle coincide.

[0057] a4 is the right third of the car's center console. A1 is when a4 overlaps with the lane line, outputting a distance of 40 cm between the vehicle body and the lane line, which is the right safety point of the vehicle.

[0058] a5 is the position of the left support column of the car. A1 is the state where a5 overlaps with the lane line angle of the scene point, and the output is that the car body is 40 cm away from the lane line, which is the left safety point of the vehicle. The position of the left support column of the car is specifically the support column between the front windshield and the center console.

[0059] S3. Analyze scene points using the driving perspective recorded by the head-mounted camera, and mark the scene points and vehicle body points in the database in the driving perspective video;

[0060] The lane line recognition algorithm used in S3 to analyze scene points includes the following steps:

[0061] S3.1. After selecting a frame of road image from the driving perspective video, obtain a road grayscale image;

[0062] S3.2. Perform image segmentation on the grayscale image: Select a grayscale value and use it as the boundary to divide the image pixel grayscale values ​​into two categories, large and small. Calculate the number of pixels in the two categories and their average values. Then calculate their inter-class variance. The grayscale value corresponding to the maximum inter-class variance is taken as the segmentation threshold.

[0063] S3.3. Use structural elements to detect the segmented image, use morphological gradients for edge detection, filter out some noise points, and detect two parallel straight lines as lane lines;

[0064] By determining the grayscale image in one frame of the road image and then segmenting the grayscale image into a threshold, since lane lines are two parallel straight lines, when performing edge detection through morphological gradient, two parallel straight lines can be detected as lane lines, which enables the head-mounted camera to recognize lane lines. Lane lines are also scene points. Based on the clustering of points into lines, it can be seen that lane lines are composed of multiple scene points, which facilitates the subsequent overlap judgment with vehicle body points.

[0065] Specifically, the principle of the segmentation threshold is further disclosed, so that the calculation formula of S3.2 is:

[0066] w(i)=n1(u)n2(i)[v1(i)-v2(i)] 2

[0067] thre=ArgMax(w(i))

[0068] Among them, w(i) is the grayscale value of the image pixel, n1(i) is the number of pixels with a grayscale value less than i, n2(i) is the number of pixels with a grayscale value greater than or equal to i, v1(i) and v2(i) are the average values ​​of the number of pixels, respectively, and thre is the subscript corresponding to when w(i) takes the maximum value, that is, the segmentation threshold. After obtaining the road grayscale image, in order to separate the target (lane line) and the background (non-lane line), image segmentation processing is required. The key to image segmentation is to select an appropriate threshold. Improper threshold selection will cause image details to be blurred or lost, etc. Therefore, it is more accurate to determine the segmentation threshold through the above calculation formula.

[0069] S4. Input the scene points into the database and calculate the overlap between the scene points and the vehicle body points. When the scene points and the vehicle body points overlap, the vehicle body is in different positions, which may be a safe position or a line-crossing position. The overlap of multiple points is judged to improve the accuracy of the alignment and accurately output the direction of the vehicle body deviation, thereby improving the adjustment effect of the subsequent driving students;

[0070] In S4, the image coordinate comparison algorithm is used to calculate the coincidence of scene points and vehicle body points. Figure 4 As shown, the following steps are included:

[0071] Call up a frame from the driving perspective video, divide the image into multiple equal regions, establish coordinate axes based on the image size, and define the coordinate point set of the vehicle body point line as: (X, Y) = [(X1, Y1), (X2, Y2), ...], and the coordinate point set of the scene point as: (x, y) = [(x1, y1), (x2, y2), ...], if (X, Y) ≈ (x, y), it represents the scene point and vehicle body point;

[0072] Working principle: Since the coordinate points of multiple vehicle point lines are connected into a straight line, and the coordinate points of multiple scene points are connected into a straight line, if the coordinate point set of the vehicle point line and the coordinate point set of the scene point are the same when taking integers, it means that the vehicle point coincides with the scene point, that is, the vehicle point coincides with the lane line.

[0073] S5. Output the overlapping status of each point and broadcast it. At the same time, transmit the driving perspective video with point marks to the instructor through the external display module. The driving perspective recorded by the head-mounted camera can not only realize the voice broadcast reminder when the driving student crosses the line, so as to avoid the driving student not being able to find out the problem and correct it in time, but also cultivate the student's driving sense to be safer. In addition, the instructor can judge whether the current student has the problem of crossing the line through the driving perspective, so that the student can quickly understand the driving defects, improve the pertinence of subsequent teaching, and benefit the teaching quality.

[0074] A second object of the present invention is to provide a vehicle body point and scene point coincidence determination system, including any one of the above vehicle body point and scene point coincidence determination methods, including a database 100, a video acquisition unit 200, a scene point analysis unit 300, a coincidence calculation unit 400 and a signal output unit 500, wherein the output end of the database 100 is connected to the output ends of the video acquisition unit 200 and the coincidence calculation unit 400 respectively, the output end of the video acquisition unit 200 is connected to the output ends of the scene point analysis unit 300 and the signal output unit 500, the output end of the scene point analysis unit 300 is connected to the output end of the coincidence calculation unit 400, and the output end of the coincidence calculation unit 400 is connected to the output end of the signal output unit 500, wherein:

[0075] The database 100 is used to store the configured vehicle body points and the driving perspective data of the camera acquisition unit 200;

[0076] The camera acquisition unit 200 is used to record driving perspective video by wearing a head-mounted camera;

[0077] The scene point analysis unit 300 is used to analyze scene points through the driving perspective video stored in the database 100, and mark the scene points and vehicle body points in the database in the driving perspective video;

[0078] The overlap calculation unit 400 is used to calculate the overlap between the scene points of the scene point analysis unit 300 and the vehicle body points stored in the database 100;

[0079] The signal output unit 500 is used to input the overlap status into the database 100, output the safety distance data for broadcast, and transmit the driving perspective video with point marks to the instructor through the external display module;

[0080] The present invention configures vehicle body point positions and coincidence status data in a database 100, so that when the camera acquisition unit 200 records a video from a driving perspective, the scene point positions are first determined according to the scene point analysis unit 300, and then the scene point positions are input into the database 100. The coincidence status of the scene point positions and the vehicle body point positions is calculated by the coincidence calculation unit 400. At the same time, the safety distance data of the coincidence status data is output through the signal output unit 500 for broadcasting, so as to remind the driving trainee and facilitate correction reminders when the line is crossed or the points are not aligned, thereby improving the efficiency of driving training and improving safety.

[0081] Specifically, the video acquisition unit 200 includes a head-mounted camera and a driving perspective recording module. The head-mounted camera is used to shoot driving videos from the driving perspective, and the driving perspective recording module is used to temporarily store driving videos to ensure the integrity of the video acquisition.

[0082] In addition, the output end of the signal output unit 510 is also connected to an external display module 510 and a voice broadcast module 520. The external display module 510 is used to transmit the driving perspective video to the display screen for display, so that the coach can observe the driving perspective to teach the driving student. The voice broadcast module 520 is used to give a voice broadcast reminder when the driving student crosses the line, so that the driving student can be reminded by voice when crossing the line, so as to avoid the driving student being unable to find his own problems and correct them in time, and at the same time cultivate the student's sense of safety in the car. The coach can judge whether the current student has the problem of crossing the line through the driving perspective, so that the student can quickly understand the driving defects, improve the targetedness of subsequent teaching, and benefit the teaching quality.

[0083] The above shows and describes the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions are merely preferred examples of the present invention and are not intended to limit the present invention. Various changes and improvements may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and improvements fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for determining the coincidence of vehicle body points and scene points, characterized in that: The steps include: S1. Wear a head-mounted camera on the driver's head to record the driving perspective video; S2. Embed a point database in the head-mounted camera and store the vehicle body points in the database; S3. Analyze scene points using the driving perspective recorded by the head-mounted camera, and mark the scene points and vehicle body points in the database in the driving perspective video; S4. Input the scene points into the database and calculate the overlap between the scene points and the vehicle body points; S5: Output the overlapping status of each point and broadcast it, and transmit the driving perspective video with the point mark to the instructor through the external display module; the head-mounted camera in S1 is fixed on the driver's head with a strap, and the camera is a sports camera; the database in S2 uses a chain memory, and its expression is: A=[(a1, A1), (a2, A2),..., (ak, Ak)], Wherein, A represents a set of vehicle body points configured in the database, a1, a2, ..., ak represent the positions of the vehicle body points, A1, A2, ..., Ak represent the state after a1 overlaps with the scene points, and k is the number of points; the calculation of the overlap between the scene points and the vehicle body points in S4 adopts an image coordinate comparison algorithm, including the following steps: Select a frame from the driving perspective video and divide the image into multiple equal regions. Establish coordinate axes based on the image size and define the coordinate point set of the vehicle body line as (X, Y) = [(X1, Y1), (X2, Y2), ...] and the coordinate point set of the scene point as (x, y) = [(x1, y1), (x2, y2), ...] . If (X, Y) ≈ (x, y), the scene point and the vehicle body point overlap.

2. The method for determining coincidence between vehicle body points and scene points according to claim 1, characterized in that: The lane line recognition algorithm is used to analyze the scene points in S3, including the following steps: S3.

1. After selecting a frame of road image from the driving perspective video, obtain a road grayscale image; S3.

2. Perform image segmentation on the grayscale image: Select a grayscale value and use it as the boundary to divide the image pixel grayscale values ​​into two categories, large and small. Calculate the number of pixels in the two categories and their average values. Then calculate their inter-class variance. The grayscale value corresponding to the maximum inter-class variance is taken as the segmentation threshold. S3.

3. Use structural elements to detect the segmented image, use morphological gradients for edge detection, filter out some noise points, and detect two parallel straight lines as lane lines.

3. The method for determining coincidence between vehicle body points and scene points according to claim 2, characterized in that: The calculation formula of S3.2 is: w(i)=n_1 (i) n_2 (i) [v_1 (i)-v_2 (i)]^2, thre=ArgMax(w(i)), Where w(i) is the grayscale value of the image pixel, n_1(i) is the number of pixels with a grayscale value less than i, n_2(i) is the number of pixels with a grayscale value greater than or equal to i, v_1(i) and v_2(i) are the average values ​​of the number of pixels, and thre is the subscript corresponding to the maximum value of w(i), i.e., the segmentation threshold.

4. A system for determining the coincidence of vehicle body points and scene points, comprising the method for determining the coincidence of vehicle body points and scene points according to any one of claims 1 to 3, characterized in that: The system comprises a database (100), a video acquisition unit (200), a scene point analysis unit (300), a coincidence calculation unit (400) and a signal output unit (500), wherein the output end of the database (100) is connected to the output ends of the video acquisition unit (200) and the coincidence calculation unit (400), respectively; the output end of the video acquisition unit (200) is connected to the output ends of the scene point analysis unit (300) and the signal output unit (500); the output end of the scene point analysis unit (300) is connected to the output end of the coincidence calculation unit (400); and the output end of the coincidence calculation unit (400) is connected to the output end of the signal output unit (500), wherein: The database (100) is used to store configured vehicle body points and driving angle data of the camera acquisition unit (200); The video acquisition unit (200) is used to record driving perspective video by wearing a head-mounted camera; The scene point analysis unit (300) is used to analyze scene points through the driving perspective video stored in the database (100), and mark the scene points and vehicle body points in the database in the driving perspective video; The overlap calculation unit (400) is used to calculate the overlap status between the scene points of the scene point analysis unit (300) and the vehicle body points stored in the database (100); The signal output unit (500) is used to input the overlap condition into the database (100), output the safety distance data for broadcast, and simultaneously transmit the driving perspective video with point marks to the instructor via an external display module.

5. The method for determining the coincidence of vehicle body points and scene points according to claim 4, characterized in that: The video acquisition unit (200) comprises a head-mounted camera and a driving perspective recording module, wherein the head-mounted camera is used to shoot a driving video from a driving perspective, and the driving perspective recording module is used to temporarily store the driving video.

6. The method for determining the coincidence of vehicle body points and scene points according to claim 4, characterized in that: The output end of the signal output unit (510) is further connected to an external display module (510) and a voice broadcast module (520). The external display module (510) is used to transmit the driving perspective video to a display screen for display, so that the instructor can observe the driving perspective to teach the driving trainee. The voice broadcast module (520) is used to give a voice broadcast reminder when the driving trainee crosses the line.

Images