A method and system for detecting the opening and closing status of four-corner locks of a motor vehicle

Abstract

The present invention discloses a method and system for detecting the opening and closing state of a four-corner lock of a motor vehicle. The method comprises: collecting an image of a four-corner lock of a motor vehicle and dividing it into a plurality of sub-images; determining the position of a four-corner lock area in each sub-image; detecting the position of each four-corner lock opening and closing state mark and the direction of the four-corner lock; judging the switch status of the four-corner lock according to the position of the four-corner lock opening and closing state mark and the direction of the four-corner lock, if the direction of the four-corner lock is consistent with the open state mark, it indicates that the four-corner lock is open, and if the direction of the four-corner lock is consistent with the closed state mark, it indicates that the four-corner lock is closed. High-resolution two-dimensional image data of the four-corner lock is obtained by a non-contact optical imaging method, and the position of the four-corner lock area is extracted from the large image of the entire carriage by a deep learning positioning algorithm, and then the direction of the four-corner lock and the open and close characters are extracted by a deep learning algorithm in the sub-image, and the opening and closing state of the four-corner lock is judged by determining the relative position relationship between the direction of the four-corner lock and the character switch.

Description

Technical Field

[0001] The invention belongs to the field of image recognition, and in particular relates to a method and a system for detecting the open and closed states of four-corner locks of a motor vehicle. Background Art

[0002] The four-corner lock is usually located on the skirt of the train and is a safety lock for the skirt. When the lock is in the open state, the skirt can be opened. When the train is running, the four-corner lock needs to be in the locked state, otherwise it will affect the safe operation of the train. At the same time, the switch direction of the four-corner lock in different parts is inconsistent.

[0003] The existing method still relies on manual inspection to confirm whether the four-corner locks are open. The manual inspection process is that the train passes at a low speed, and a person stands beside the train (within the safety limit) to check the open and closed status of the four-corner locks with his eyes, and uses a camera flashlight to record the video at the same time. The shortcomings of manual inspection are:

[0004] 1. Visual fatigue, prone to missed things

[0005] 2. The train speed should not be too fast, otherwise the human eye cannot distinguish it and the efficiency is low.

[0006] 3. The resolution of a camera flashlight is generally low, and when people shoot with their hands held, the shooting posture may not be optimal. Summary of the invention

[0007] In view of this, the present invention provides a method and system for detecting the open and closed status of the four-corner locks of a motor vehicle, which can automatically identify and determine the open and closed status of the four-corner locks.

[0008] To solve the above technical problems, the present invention provides a method for detecting the open and closed state of a motor vehicle four-corner lock, comprising: collecting an image of the motor vehicle four-corner lock and dividing it into a plurality of sub-images; determining the position of the four-corner lock area in each sub-image; detecting the position of each four-corner lock open and closed state mark and the direction of the four-corner lock; judging the switch status of the four-corner lock according to the position of the four-corner lock open and closed state mark and the direction of the four-corner lock, if the direction of the four-corner lock is consistent with the open state mark, it indicates that the four-corner lock is open, and if the direction of the four-corner lock is consistent with the closed state mark, it indicates that the four-corner lock is closed.

[0009] As an improvement, the collecting of the four-corner lock image of the motor vehicle and dividing it into a plurality of sub-images includes: setting a sliding window, the sliding window slides from the beginning of the image to the end of the image, and each time the sliding window slides, the image framed in the sliding window is cut into a sub-image.

[0010] As a preference, the sliding window has the same height as the image, a width of L, and a sliding step length of L / 2.

[0011] As an improvement, after determining the position of the four-corner lock area in each sub-image, before detecting the position of the opening and closing status mark of each four-corner lock and the direction of the four-corner lock, find the sub-image that repeatedly records the same four-corner lock, and retain one of each group of repeated sub-images.

[0012] As an improvement, the method for finding sub-images that repeatedly record the same four-corner lock includes: converting the coordinates of all four-corner lock areas into the coordinate system of the entire four-corner lock image; dividing the absolute value of the intersection of the coordinates of the four-corner lock areas in adjacent sub-images by the absolute value of the union of the coordinates of the four-corner lock areas in adjacent sub-images. If it is greater than a threshold, it is considered that the two repeatedly record the same four-corner lock.

[0013] As an improvement, the method for converting the coordinates of all four-corner lock areas to the coordinate system of the entire four-corner lock image includes: marking the regional coordinates (x, y, w, h) for the regional position of each four-corner lock, wherein y is the coordinate in the train running direction, x is the coordinate in the train height direction, w is the width of the four-corner lock area, and h is the height of the four-corner lock area;

[0014] Using the formula

[0015] X m =x m

[0016] Y m =L1 / 2×(M-1)+y m

[0017] W m =w m

[0018] H m =h m

[0019] Convert the coordinates of the four-corner lock region in the mth sub-image to the coordinate system of the whole vehicle image before segmentation, where x m is the train height coordinate of the four-corner lock area in the mth sub-image, ym is the train running direction coordinate of the four-corner lock area in the mth sub-image, h m is the height of the four-corner lock area in the mth sub-image, w m is the width of the four-corner lock area in the mth sub-image, X m Y is the height coordinate of the four-corner lock area in the whole vehicle image. m H is the coordinate of the four-corner lock area in the train moving direction in the whole vehicle image, m is the height of the four-corner lock area in the vehicle image, W m is the width of the corner lock area in the whole vehicle image.

[0020] As a preferred embodiment, a confidence c is assigned to each sub-image, and the sub-image with the highest confidence is retained in each group of repeated sub-images.

[0021] As an improvement, if the position of the open / closed status mark of the four-corner lock fails to be detected, the switch status of the four-corner lock is determined by the orientation of the four-corner lock and the position of the open / closed status mark of the four-corner lock in the standard database.

[0022] As an improvement, the method for generating a standard database is as follows: recording a number of status information of each four-corner lock, wherein the status information includes classification information of an open state marking position, classification information of a closed state marking position, and classification information of a four-corner lock orientation; statistics are performed on the classification information in all the status information recorded for each four-corner lock according to its classification, and the classification information with the highest frequency of occurrence is recorded as standard classification information.

[0023] The present invention also provides a motor vehicle four-corner lock open and closed state detection system, comprising: an image acquisition module for acquiring images of the motor vehicle four-corner locks; and a four-corner lock state detection module for detecting the open and closed state of the four-corner locks in the acquired images.

[0024] As an improvement, the four-corner lock state detection module includes: a four-corner lock area positioning module, used to determine the four-corner lock area position; a four-corner lock posture detection module, used to detect the four-corner lock direction; a marking detection module, used to detect the four-corner lock open and closed state marking; a four-corner lock state judgment module, used to judge the open and closed state of the four-corner lock based on the detection results of the four-corner lock posture detection module and the marking detection module.

[0025] As an improvement, the four-corner lock state detection module further includes a sub-image segmentation module for segmenting the collected image information into a plurality of sub-images.

[0026] As an improvement, it also includes: a standard data generation module, which is used to count several status information of each four-corner lock, and count the classification information in all the status information recorded for each four-corner lock according to its classification, and record the classification information with the highest frequency as standard classification information; when the four-corner lock status detection module cannot detect the open and closed status indication of the four-corner lock or the direction of the four-corner lock, the standard classification information is called for judgment.

[0027] The benefits of the present invention are: high-resolution two-dimensional image data of the four-corner lock is obtained through a non-contact optical imaging method, and the position of the four-corner lock area (including the status characters "open" and "closed") is extracted from the large picture of the entire carriage through a deep learning positioning algorithm, and then the direction of the four-corner lock and the open and close characters are extracted in the sub-picture (the sub-picture means that a single picture contains only one four-corner lock) by using a deep learning algorithm. By determining the relative position relationship between the direction of the four-corner lock and the character switch, the open and closed state of the four-corner lock is determined. At the same time, for the case where the characters are unclear or the paint is peeling, the corresponding algorithm logic is designed to ensure that the state of the four-corner lock can still be determined in these cases, so that the detection accuracy of the system is higher and the system adaptability is better.

[0028] 1. Suitable for non-contact measurement without damaging the target.

[0029] 2. Algorithm intelligent detection, fully automated, improves detection efficiency.

[0030] 3. High universality and easy to promote and apply.

[0031] 4. High-resolution cameras can adapt to fast vehicle speeds, and the image data is also conducive to archiving.

[0032] 5. Based on artificial intelligence algorithm and data statistical algorithm, the algorithm has good adaptability and high detection accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 It is a schematic diagram of the four-corner lock in the present invention.

[0034] Figure 2 It is a flow chart of the present invention.

[0035] Figure 3 This is a flow chart for generating the standard database in the present invention.

[0036] Figure 4 It is a structural principle diagram of the present invention. DETAILED DESCRIPTION

[0037] In order to enable those skilled in the art to better understand the technical solution of the present invention, the present invention is further described in detail below in conjunction with specific implementation methods.

[0038] Figure 1It is a schematic diagram of a four-corner lock, including a four-corner lock body and an open / closed status indication. In the present invention, a triangular arrow direction mark is provided on the four-corner lock body, and the open / closed status indication includes the characters of "open" and "closed". The open / closed status indication is arranged on the circumference outside the four-corner lock body and is generally located at the top, bottom, left, and right of the four-corner lock body. In the present invention, the top of the four-corner lock body is defined as position 0, the bottom is defined as position 2, the left is defined as position 3, and the right is defined as position 1. Of course, each four-corner lock has only two states: open and closed, so the open / closed status indication of each four-corner lock only occupies two positions, and the two positions are generally continuous. As shown in FIG. Figure 1 As shown, the open state mark is located above the four-corner lock body, the closed state mark is located on the right side of the four-corner lock, and the direction mark pointing to the right indicates that the state of the four-corner lock is closed.

[0039] Although the four-corner locks are in various forms, the opening and closing status indication principles are similar to the four-corner locks listed in the present invention, so the present invention is not limited to the four-corner lock types listed above.

[0040] like Figure 2 As shown in the flowchart, the present invention provides a method for detecting the open and closed state of the four-corner locks of a motor vehicle, comprising:

[0041] S1 collects the image of the four-corner lock of the train and divides it into several sub-images; the collection process is that the image collection module is fixed, and when the train passes, the image collection module starts to collect images and saves them to the server. In this embodiment, the sub-image is divided by a sliding window method, that is, a sliding window is set, and the sliding window slides from the beginning of the image to the end of the image. Each time it slides (except the first sub-image), the image framed in the sliding window is cut into a sub-image. The sliding window is the same height as the collected image, the width is L, and the sliding step length is L / 2.

[0042] S2 determines the position of the four-corner lock area in each sub-image; in this embodiment, the AI ​​model YOLOV3 is used to determine the position of the four-corner lock area. The definition of the four-corner lock is used to train the model A at the beginning of the specific implementation method. After the model training is completed, the position of the four-corner lock area can be automatically located by inputting the sub-image model A. Of course, other AI algorithms or other forms can also be used to determine the position of the four-corner lock. There is no limitation in the present invention, as long as the position of the four-corner lock area can be determined.

[0043] S3 finds the subgraphs that record the same four-corner lock repeatedly, and retains one of each set of repeated subgraphs. The purpose of this step is to remove duplicates and avoid wasting system resources and generating erroneous data due to repeated identification. The specific steps include:

[0044] S31 converts the coordinates of all four-corner lock areas to the coordinate system of the entire four-corner lock image, specifically:

[0045] S311 marks the regional coordinates (x, y, w, h) for the regional position of each four-corner lock, where y is the coordinate in the train running direction, x is the coordinate in the train height direction, w is the width of the four-corner lock area, and h is the height of the four-corner lock area; at the same time, in this step, while marking the regional coordinates, the confidence c of the four-corner lock area is given.

[0046] S312 uses the formula

[0047] X m =x m

[0048] Y m =L1 / 2×(M-1)+y m

[0049] W m =w m

[0050] H m =h m

[0051] Convert the coordinates of the four-corner lock region in the mth sub-image to the coordinate system of the whole vehicle image before segmentation, where x m is the train height coordinate of the four-corner lock area in the mth sub-graph, y m h is the train running direction coordinate of the four-corner lock area in the mth sub-graph, m is the height of the four-corner lock area in the mth sub-image, w m is the width of the four-corner lock area in the mth sub-image, X m Y is the height coordinate of the four-corner lock area in the whole vehicle image. m H is the coordinate of the four-corner lock area in the train moving direction in the whole vehicle image, m is the height of the four-corner lock area in the vehicle image, W m is the width of the corner lock area in the whole vehicle image.

[0052] S32 divides the absolute value of the intersection of the coordinates of the four-corner lock regions in the adjacent sub-graphs by the absolute value of the union of the coordinates of the four-corner lock regions in the adjacent sub-graphs. If the absolute value is greater than a threshold, it is considered that the two duplicate records of the same four-corner lock.

[0053] Determine the coordinate relationship of the four-corner locks in two adjacent images in turn. Since the four-corner locks have been restored to the same coordinate system, if there is the same four-corner lock in the two sub-images, their coordinate areas will overlap.

[0054] For example, the Mth Figure 4 The angle lock coordinates are LOCK_m(X m , Y m , W m , H m), confidence C_m, and the four corner lock coordinates in the M+1th image are LOCK_m+1(X m+1 , Y m+1 , W m+1 , H m+1 ), confidence C_m+1. To determine whether LOCK_m and LOCK_m+1 are the same four-corner lock, the intersection and union ratio of LOCK_m and LOCK_m+1 is calculated. The algorithm sets a threshold th1. When the intersection and union ratio of two four-corner locks is greater than th1, it means that the two areas have a good overlap and belong to the same four-corner lock. The formula is as follows:

[0055] Where IOU is the intersection-over-union ratio.

[0056] S33 Since each four-corner lock has a confidence level, the object to be retained is selected based on the confidence level. If C_m+1>C_m, then LOCK_m+1 is retained, otherwise LOCK_m is retained.

[0057] S4 detects the position of the opening and closing status mark of each four-corner lock and the direction of the four-corner lock; in this embodiment, this step is still implemented using an AI algorithm, namely, an AI model unet for identifying the position of the opening and closing status mark of the four-corner lock, and an AI model YOLOV3 for identifying the direction of the four-corner lock, which are model B and model C respectively. Similarly, other AI algorithms or other forms can also be used to detect the position of the opening and closing status mark of each four-corner lock and the direction of the four-corner lock, which is not limited in the present invention.

[0058] The deduplicated sub-graphs are input into models B and C. Model B will output the “open” and “closed” position information of each quadrilateral lock sub-graph, open(x1, y1, w1, h1) and close(x2, y2, w2, h2). Model C will output the area information of the inner triangle of the quadrilateral lock, triangle(x3, y3, w3, h3) and direction.

[0059] S5 determines the switch status of the four-corner lock according to the position of the open and closed status mark of the four-corner lock and the direction of the four-corner lock. If the direction of the four-corner lock is consistent with the open status mark, it means that the four-corner lock is open; if the direction of the four-corner lock is consistent with the closed status mark, it means that the four-corner lock is closed.

[0060] Determine the positions of open and close in the triangle (up / down / left / right). If the open position is consistent with the direction position, the four-corner lock is open; if the close position is consistent with the direction position, the four-corner lock is closed.

[0061] In addition, due to various reasons such as the wear of the four-corner lock opening and closing status mark, Model B and Model C cannot feedback the four-corner lock opening and closing status mark or the four-corner lock direction. In order to ensure the robustness of the system, the present invention also has:

[0062] If the position of the four-corner lock opening and closing status mark fails to be detected in S6, the opening and closing status of the four-corner lock is determined by the orientation of the four-corner lock and the position of the four-corner lock opening and closing status mark in the standard database.

[0063] The method for generating a standard database is:

[0064] S61 records a number of status information of each four-corner lock, wherein the status information includes classification information of the open state marking position, classification information of the closed state marking position, and classification information of the four-corner lock orientation;

[0065] S62 counts the classification information in all the status information of each four-corner lock according to its classification, and records the classification information with the highest frequency as standard classification information.

[0066] like Figure 3 As shown, according to the orientation information of each four-corner lock and the characters "open" and "close" obtained in the above steps, N data are recorded. Then each four-corner lock obtains N state information {state1, state2, ... stateI, stateN}. Among them, stateN = {open_direction_N (open direction), close_direction_N (closed direction), triangle_direction_N (four-corner lock direction)}.

[0067] The possible values ​​of open_direction_N, close_direction_N, and triangle_direction_N are {1 (up), 2 (down), 3 (left), and 4 (right)}. Therefore, by using the probability statistics method, the number of the largest values ​​among the N open_directions is counted, and the value is recorded as the normal position information of the "open" character and stored in the database. Similarly, the position information of the "close" character and the four corner lock directions can be calculated.

[0068] For example, if open_direction = {3 3 1 3 3 3 1 3 null 3}, 3 appears 7 times, 1 appears 2 times, and null indicates that the detection failed. Therefore, it can be concluded that 3 appears most frequently, and the position state of "open" is left.

[0069] Therefore, when model B or model C cannot be output, and the standard database has been generated, it is determined whether the "open" orientation in the standard database is consistent with the currently detected four-corner lock orientation. If the standard database has not been generated, the sub-area map of the four-corner lock is directly uploaded to the information display platform for manual verification.

[0070] like Figure 4 As shown, the present invention also provides a motor vehicle four-corner lock open and closed state detection system, comprising:

[0071] The image acquisition module is used to acquire images of the four corner locks of the motor vehicle; in this embodiment, the image acquisition module uses a linear array CCD to acquire two-dimensional images. Of course, other types of image acquisition modules can also be selected.

[0072] The four-corner lock state detection module is used to detect the open and closed states of the four-corner locks in the collected image, specifically including:

[0073] A sub-image segmentation module is used to segment the collected image information into several sub-images;

[0074] Four-corner lock area positioning module, used to determine the position of the four-corner lock area;

[0075] Four-corner lock posture detection module, used to detect the orientation of the four-corner lock;

[0076] Marking detection module, used to detect the opening and closing status markings of the four corner locks;

[0077] The four-corner lock state judgment module is used to judge the open and closed state of the four-corner lock according to the detection results of the four-corner lock posture detection module and the marking detection module.

[0078] The four-corner lock state detection module receives image data from the image acquisition module. Generally, standard EMUs are divided into 8-carriage and 16-carriage. The length of the 8-carriage vehicle is about 200 meters. Therefore, the image of the entire vehicle is very long and cannot be loaded into the algorithm for detection at one time. Therefore, the image is cut into small images with overlapping areas by sliding windows, and the deep learning target detection network-YOLOV5 is used on each small image to detect the position of the four-corner lock in the image. Since there is an intersection between two adjacent windows, a four-corner lock screening algorithm is added, and only one detection box is retained at the same position.

[0079] After obtaining the position of the four-corner lock, extract the sub-region map containing the four-corner lock and the open and closed status mark. On the sub-region map (the four-corner lock and the characters account for a large proportion of the sub-region map, solving the small target problem), use the deep learning target detection network-yolov3 to detect the position of the "open" character and the position of the "closed" character. In the sub-region area, use the deep learning semantic segmentation network-unet to segment the inner triangle of the four-corner lock and determine the direction of the triangle.

[0080] Based on the detected orientation of the four-corner lock and the position of the switch, the switch state of the four-corner lock is determined.

[0081] The standard data generation module is used to count several status information of each four-corner lock, and count the classification information in all the status information recorded for each four-corner lock according to its classification, and record the classification information with the highest frequency as standard classification information; when the four-corner lock status detection module cannot detect the open and closed status indication of the four-corner lock or the direction of the four-corner lock, the standard classification information is called for judgment.

[0082] If the characters are incomplete or the triangle indicating the orientation is unclear, resulting in the inability of the marking detection module and the four-corner lock posture detection module to effectively detect, the four-corner lock status detected can be determined by comparing it with the posture information recorded in the standard database. After determining the four-corner lock status, the system will upload the results to the information platform for customers to view.

[0083] Since the automatic generation of standard data requires N rounds of data, if before the standard data is automatically generated, a single detection has incomplete characters that make it impossible to determine the detection status, in the present invention, the system will automatically push the sub-area map to the reviewer, and the reviewer will confirm the image that the system cannot determine through manual inspection, thereby ensuring that the system does not miss any detection.

[0084] The above are only preferred embodiments of the present invention. It should be noted that the above preferred embodiments should not be regarded as limiting the present invention, and the protection scope of the present invention should be based on the scope defined by the claims. For ordinary technicians in this technical field, several improvements and modifications can be made without departing from the spirit and scope of the present invention, and these improvements and modifications should also be regarded as the protection scope of the present invention.

Claims

1. A method for detecting the open and closed state of the four-corner locks of a motor vehicle, characterized in that include: Collect the four-corner lock images of the train and divide them into several sub-images; Determine the location of the four corner lock areas in each sub-image; Detect the position of the open and closed status indicator of each corner lock and the direction of the four corner locks; The switch status of the four-corner lock is determined according to the position of the open and closed status indication of the four-corner lock and the direction of the four-corner lock. If the direction of the four-corner lock is consistent with the open status indication, it means that the four-corner lock is open. If the direction of the four-corner lock is consistent with the closed status indication, it means that the four-corner lock is closed.

2. A method for detecting the open and closed state of the four-corner locks of a motor vehicle according to claim 1, characterized in that The collecting of the four-corner lock image of the motor vehicle and dividing it into a plurality of sub-images comprises: A sliding window is set, and the sliding window slides from the beginning of the image to the end of the image. Each time the sliding window slides, the image selected in the sliding window is cut into a sub-image.

3. A method for detecting the open and closed state of the four-corner locks of a motor vehicle according to claim 2, characterized in that: The sliding window has the same height as the image, a width of L, and a sliding step length of L / 2.

4. A method for detecting the open and closed state of the four-corner locks of a motor vehicle according to claim 1, characterized in that After determining the position of the four-corner lock area in each sub-image, before detecting the position of the opening and closing state mark of each four-corner lock and the direction of the four-corner lock, find the sub-image that repeatedly records the same four-corner lock, and retain one of each group of repeated sub-images.

5. A method for detecting the open and closed state of the four-corner locks of a motor vehicle according to claim 4, characterized in that The method for finding the subgraph that repeatedly records the same four-corner lock includes: Convert the coordinates of all four-corner lock areas to the coordinate system of the entire four-corner lock image; The absolute value of the intersection of the coordinates of the four-corner lock areas in the adjacent sub-graphs is divided by the absolute value of the union of the coordinates of the four-corner lock areas in the adjacent sub-graphs. If it is greater than the threshold, it is considered that the two duplicate records of the same four-corner lock.

6. A method for detecting the open and closed state of the four-corner locks of a motor vehicle according to claim 5, characterized in that The method for converting the coordinates of all four-corner lock areas to the coordinate system of the entire four-corner lock image includes: Mark the regional coordinates (x, y, w, h) for the regional position of each four-corner lock, where y is the coordinate in the train running direction, x is the coordinate in the train height direction, w is the width of the four-corner lock area, and h is the height of the four-corner lock area; Using the formula Convert the coordinates of the four-corner lock region in the mth sub-image to the coordinate system of the whole vehicle image before segmentation, where x m is the train height coordinate of the four-corner lock area in the mth sub-graph, y m h is the train running direction coordinate of the four-corner lock area in the mth sub-graph, m is the height of the four-corner lock area in the mth sub-image, w m is the width of the four-corner lock area in the mth sub-image, X m Y is the height coordinate of the four-corner lock area in the whole vehicle image. m H is the coordinate of the four-corner lock area in the train moving direction in the whole vehicle image, m is the height of the four-corner lock area in the vehicle image, W m is the width of the corner lock area in the whole vehicle image.

7. A method for detecting the open and closed state of the four-corner locks of a motor vehicle according to claim 4, characterized in that: Assign a confidence score c to each sub-image, and retain the sub-image with the highest confidence score in each set of repeated sub-images.

8. A method for detecting the open and closed state of the four-corner locks of a motor vehicle according to claim 1, characterized in that: If the position of the four-corner lock opening and closing status mark fails to be detected, the opening and closing status of the four-corner lock is determined by the orientation of the four-corner lock and the position of the four-corner lock opening and closing status mark in the standard database.

9. A method for detecting the open and closed state of the four-corner locks of a motor vehicle according to claim 8, characterized in that The method for generating a standard database is: Recording a number of status information of each four-corner lock, the status information including classification information of the open state marking position, classification information of the closed state marking position, and classification information of the four-corner lock orientation; The classified information in all the status information recorded for each four-corner lock is counted according to its classification, and the classified information with the highest frequency of occurrence is recorded as the standard classified information.

10. A system for detecting the open and closed state of the four-corner locks of a motor vehicle, characterized in that include: Image acquisition module, used to collect images of the four-corner locks of the motor vehicle; A four-corner lock state detection module is used to detect the open and closed states of the four-corner locks in the collected image; The four-corner lock state detection module includes: A sub-image segmentation module is used to segment the collected image information into several sub-images; Four-corner lock area positioning module, used to determine the position of the four-corner lock area; Four-corner lock posture detection module, used to detect the orientation of the four-corner lock; Marking detection module, used to detect the opening and closing status markings of the four corner locks; The four-corner lock state judgment module is used to judge the open and closed state of the four-corner lock according to the detection results of the four-corner lock posture detection module and the marking detection module. If the direction of the four-corner lock is consistent with the open state mark, it means that the four-corner lock is open; if the direction of the four-corner lock is consistent with the closed state mark, it means that the four-corner lock is closed.

11. A motor vehicle four-corner lock open and closed state detection system according to claim 10, characterized in that Also includes: The standard data generation module is used to count several status information of each four-corner lock, and count the classification information in all the status information recorded for each four-corner lock according to its classification, and record the classification information with the highest frequency as standard classification information; when the four-corner lock status detection module cannot detect the open and closed status indication of the four-corner lock or the direction of the four-corner lock, the standard classification information is called for judgment.

Images