Gap single-side light bar type and gap type identification method based on machine vision

Abstract

The invention discloses a gap unilateral light bar type identification method based on machine vision, and the method comprises the steps: 1) carrying out the light bar center extraction of a gap contour light bar image, and obtaining a light bar center point set which is orderly arranged between the starting end and the tail end of a unilateral light bar; 2) converting data in the point set to obtain a light plane coordinate system, and recording the light plane coordinate system as a point cloud set; 3) selecting a light bar inflection point in the point cloud set and calculating the distance from the light bar inflection point to a straight line where the starting end point and the tail end point are located, and if the distance is smaller than a threshold value, determining that the light bar is a one-section light bar; otherwise, performing the step 4); 4) searching a second inflection point which divides points between the inflection point and the tail end point into an arc point and a straight line point; if the number of the straight line points is larger than a preset threshold value, the light bar is a three-section type light bar, and otherwise, the light bar is a two-section type light bar. The method achieves the automatic recognition of the type of the single-side contour of the gap, effectively prevents the intervention of human subjective factors, and improves the quality control capability.

Description

Technical Field

[0001] This invention relates to the field of visual inspection, specifically to a machine vision-based method for identifying the type of single-sided light stripe in a gap and the gap type. Background Technology

[0002] In the field of automotive quality control, surface clearance difference is a key indicator for measuring the quality of vehicle body matching, directly affecting the vehicle's water leakage prevention performance and noise suppression performance. Currently, mainstream automobile manufacturers are gradually adopting non-contact visual measurement to replace manual measurement for body surface clearance difference. This visual measurement method not only has a high degree of automation, measurement efficiency, and measurement accuracy, but also allows for the summarization and analysis of measurement data. Therefore, visual measurement is the general trend in body surface clearance difference measurement.

[0003] Currently, common visual measurement methods for gap differences involve projecting structured light onto the surfaces on both sides of the gap to be measured, capturing images of the structured light with a camera, and analyzing the captured images to obtain a gap contour point cloud. Based on a preset gap evaluation method, the gap contour point cloud is further calculated to obtain the gap measurement result. However, before calculating the gap contour point cloud, the type of gap contour light stripe needs to be manually selected to match the corresponding preset gap evaluation method. This reduces the practicality and automation of the visual measurement system to some extent. Furthermore, the introduction of human factors can lead to inconsistent evaluation standards.

[0004] In practical applications, there are many types of gaps on the vehicle body, making it difficult to automatically identify the types of light stripes in the gaps. However, in the gap images captured by the camera, the light stripe types of the single-sided contour (the body structure corresponding to the left / right side) that constitutes the gap only include the following three types:

[0005] A single-segment light strip, such as Figure 1 As shown in A, this generally occurs in the edge cutting process of parts. Its contour point cloud can only be collected from the outer surface of the part, and the edge point cloud at the gap shows a state of almost or completely missing, with only a straight line point cloud.

[0006] Two-segment light strip, such as Figure 1 As shown in B, it generally appears in the edge-wrapping process of parts. Its contour point cloud can be collected from the outer surface of the part and the edge, and can be divided into a straight line point cloud and a circular arc point cloud.

[0007] Three-segment light strip, such as Figure 1 As shown in C, it generally appears in the flanging process of parts. Its contour point cloud can collect the outer surface of the part and the flanging. Based on the two-segment light strip point cloud, an inclined part gap inner surface point cloud is added, which can be divided into an outer straight line point cloud, an arc point cloud and an inner straight line point cloud.

[0008] Various gap profile light stripe types are formed by freely combining the above three single-sided profile light stripe types; based on this feature, how to achieve automatic recognition of single-sided profile light stripe types is the key to achieving automatic recognition of gap profile light stripes. Summary of the Invention

[0009] To address the aforementioned technical problems, this invention provides a machine vision-based method for identifying the type of light stripe on one side of a gap and the gap type. This method can automatically identify the type of contour on one side of a gap, avoiding the intervention of human factors and thus achieving complete automation of gap visual measurement.

[0010] Therefore, the technical solution of the present invention is as follows:

[0011] A machine vision-based method for identifying single-sided light stripes in gaps, where the light stripe is perpendicular to the gap extension direction during testing; includes the following steps:

[0012] 1) Extract the center of the light stripe from the image of the light stripe contour to obtain the center point set A of the light stripe; denote the position of the light stripe close to the gap as the end and the position far from the gap as the beginning; denote the center points of the light stripe arranged in an orderly manner between the beginning and end of the light stripe on one side as the point set A1;

[0013] 2) Transform the data in point set A1 from two-dimensional coordinates in the image coordinate system to three-dimensional coordinates in the camera coordinate system, and then to two-dimensional coordinates in the light plane coordinate system, denoted as point cloud set A1';

[0014] 3) Process the data in the point cloud A1', select the feature points corresponding to the inflection points of the light stripe, the starting feature point, and the ending feature point, and calculate the distance D from the inflection point of the light stripe to the straight line containing the starting and ending feature points based on these points; set a distance threshold ThD. If D < ThD, then the light stripe is a single-segment light stripe; otherwise, proceed to step 4).

[0015] 4) Find the second inflection point between the inflection point and the end feature point. Denote the feature points between the inflection point and the second inflection point as the arc segment point set C1, and denote the feature points between the second inflection point and the end feature point as the straight line segment point set D1.

[0016] Set a distance threshold ThL; perform circular arc fitting on the point cloud C1, and calculate the distance L1 from each feature point in C1 to the fitted circular arc; if L1 < ThL, then mark the feature point as an arc point.

[0017] Perform a straight line fitting on the point cloud D1, and calculate the distance L2 from each feature point in D1 to the fitted straight line; if L2 < ThL, then mark the feature point as a straight line point.

[0018] If the number of feature points marked as straight lines is greater than the number of feature points marked as arcs, and the number of feature points in point cloud D1 is greater than a preset threshold, then the light stripe is a three-segment light stripe; otherwise, it is a two-segment light stripe.

[0019] Further, step 1) involves obtaining the point set A1 as follows: Starting from the initial endpoint P1 of the light stripe, a nearest neighbor search algorithm is used in the light stripe center point set A to search for its nearest point P2 in the remaining light stripe center point set with a preset threshold. Then, P2 is removed from the remaining light stripe center point set and used as a new starting point. The search continues until the search point P1 is reached. n The process stops when no nearest point is found within a preset threshold. Then, points P1, P2, P3, ..., P... n The set of center points of the light stripe is A1; P n That is, the end point of the light stripe.

[0020] Furthermore, the method for selecting the inflection point of the light stripe in step 3) is as follows: taking the end feature point in the point cloud A1' as the starting point and performing iterative calculations towards the starting feature point of the light stripe, in the k-th iteration, a feature point with a distance of k×N points from the starting point is selected as the endpoint, where k is a positive integer and N is the preset number of points; in each iteration, the distance W from all feature points between the starting point and the endpoint to the straight line between the starting point and the endpoint is calculated one by one, and the point with the largest W in each iteration is recorded as a pre-selected inflection point; until k×N is greater than the number of feature points in A1', the starting feature point of the light stripe is used as the endpoint of this iteration to calculate the pre-selected inflection point, and the iteration ends; the point with the largest W among all pre-selected inflection points is recorded as the inflection point.

[0021] Furthermore, the method for extracting the center of the light stripe in step 1) is: Steger algorithm or grayscale centroid method.

[0022] This invention also provides a machine vision-based gap type recognition method, which uses a machine vision-based gap single-sided light stripe type recognition method to obtain the single-sided light stripe type on both sides of the gap, including sharp-edge gaps, beveled-edge gaps, rolled-edge gaps, and T-shaped gaps; the gap type is confirmed according to the following rules:

[0023] The sharp-edge gap consists of single-sided light strips on both sides of the gap, both of which are single-section light strips.

[0024] The oblique gap is a type of three-segment light strip with light strips on either side of the gap.

[0025] The rolled edge gap includes double rolled edge gap and sharp rolled edge gap; double rolled edge gap is when the light strips on both sides of the gap are of the two-segment light strip type; sharp rolled edge gap is when the light strips on one side of the gap are of the one-segment light strip type and the light strips on the other side are of the two-segment light strip type, and the supplementary judgment conditions are not met.

[0026] A T-shaped gap is defined as a gap where one side of the light bar is a single-segment light bar and the other side of the light bar is a two-segment light bar, and any supplementary judgment condition is met;

[0027] The supplementary judgment condition is:

[0028] Condition 1: Perform straight line fitting on the point cloud from the starting point to the inflection point in the point cloud set A1' corresponding to the two-segment light stripe to obtain the outer straight line of the light stripe. Draw a perpendicular line from the end feature point to the outer straight line of the light stripe to obtain the positional relationship between the one-segment light stripe and the perpendicular line. If the one-segment light stripe is divided into two parts by the perpendicular line, then it is determined that Condition 1 is satisfied; otherwise, it is not satisfied.

[0029] Condition 2: Perform straight line fitting on the point cloud from the starting feature point to the inflection point in the two single-sided light stripe point cloud sets A1' respectively, and calculate the angle between the two fitted straight lines. If the angle exceeds the preset angle threshold, then condition 2 is satisfied; otherwise, condition 2 is not satisfied.

[0030] Furthermore, the types of beveled edge gaps include double beveled edge gaps, rolled beveled edge gaps, and sharp beveled edge gaps;

[0031] The double-bevel gap type is a three-segment light bar type with single-sided light bars on both sides.

[0032] The bevel gap type is one single-sided light bar type is three-segment light bar, and the other single-sided light bar is two-segment light bar;

[0033] The sharp bevel gap type is a gap with one single-sided light bar type being a three-segment light bar, and the other single-sided light bar being a one-segment light bar.

[0034] This invention achieves the goal of automatically identifying the single-sided contour of gaps, and further, by analyzing the relationship between the single-sided contour type and the gap contour type, it achieves automatic identification of the gap contour type, improving the automation level of the gap visual measurement system, effectively preventing the intervention of human subjective factors, and processing all data using a unified evaluation standard, effectively ensuring the standardization and normalization of data processing, and avoiding human differences and errors. In addition, compared with existing methods, this invention can trace the type identification data of each gap, supporting quality traceability and problem investigation, and effectively improving quality control capabilities. Attached Figure Description

[0035] Figure 1 A schematic diagram of point cloud classification for unilateral gap contours;

[0036] Figure 2 A schematic diagram illustrating the method for selecting the inflection point of a light stripe;

[0037] Figure 3 This is a flowchart illustrating the gap type identification method for applying the single-sided light stripe type identification method of the present invention. Detailed Implementation

[0038] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and embodiments.

[0039] When structured light is projected perpendicularly onto the gap being measured, it will form two single-sided light stripe images perpendicular to the direction of gap extension, specifically including the left light stripe corresponding to the left contour of the gap and the right light stripe corresponding to the right contour of the gap.

[0040] A machine vision-based method for identifying single-sided light stripes in gaps, where the light stripe is perpendicular to the gap extension direction during testing; includes the following steps:

[0041] 1) Extract the center of the light stripe from the image of the light stripe contour to obtain the center point set A of the light stripe; denote the position of the light stripe close to the gap as the end and the position far from the gap as the beginning; denote the center points of the light stripe arranged in an orderly manner between the beginning and end of the light stripe on one side as the point set A1;

[0042] 2) Using the camera calibration parameters of the sensor that acquires the light stripe image, the data in the point set A1 is transformed from the two-dimensional coordinates of the image coordinate system to the three-dimensional coordinates of the camera coordinate system; then, using the structured light plane parameters of the sensor calibration result, the three-dimensional coordinates in the camera coordinate system are transformed to the two-dimensional coordinates in the light plane coordinate system, denoted as point cloud set A1'.

[0043] 3) Process the data in the point cloud A1', select the feature points corresponding to the inflection points of the light stripe, the starting feature point, and the ending feature point, and calculate the distance D from the inflection point of the light stripe to the straight line containing the starting and ending feature points based on these points; set a distance threshold ThD. If D < ThD, then the light stripe is a single-segment light stripe; otherwise, proceed to step 4). Preferably, the distance threshold ThD is 0.2~0.3mm.

[0044] 4) Find the second inflection point between the inflection point and the end feature point. Denote the feature points between the inflection point and the second inflection point as the arc segment point set C1, and denote the feature points between the second inflection point and the end feature point as the straight line segment point set D1.

[0045] Set a distance threshold ThL; perform circular arc fitting on the point cloud C1, and calculate the distance L1 from each feature point in C1 to the fitted circular arc; if L1 < ThL, then mark the feature point as an arc point.

[0046] Perform a straight line fitting on the point cloud D1, and calculate the distance L2 from each feature point in D1 to the fitted straight line; if L2 < ThL, then mark the feature point as a straight line point; preferably, the distance threshold ThL is 0.1~0.3mm.

[0047] If the number of feature points marked as straight lines is greater than the number of feature points marked as arcs, and the number of feature points in point cloud D1 is greater than a preset threshold, then the light stripe is a three-segment light stripe; otherwise, it is a two-segment light stripe.

[0048] In specific implementation, in order to obtain an ordered set of feature points, step 1) of obtaining the point set A1 is as follows: Starting from the starting endpoint P1 of the light stripe, the nearest neighbor search algorithm is used in the set of center points A of the light stripe to search for its nearest point P2 in the set of center points of the remaining light stripe with a preset threshold. Then, P2 is removed from the set of center points of the remaining light stripe and used as a new starting point. The nearest point P3 of P2 is searched until the search point P is reached. n The process stops when no nearest point is found within a preset threshold. Then, points P1, P2, P3, ..., P... n The set of center points of the light stripe is A1; P n That is, the end point of the light stripe.

[0049] Furthermore, in step 3), the method for selecting the inflection point of the light stripe is as follows: using the terminal feature point P in the point cloud A1'. n Starting from the initial feature point of the light stripe, perform iterative calculations. In the k-th iteration, select a feature point k×N points away from the starting point as the endpoint P. x Where k is a positive integer and N is the preset number of points; in each iteration, the starting point P is calculated one by one. n To endpoint P x All feature points between the starting point Pn and the endpoint P x The distance W along the straight line is used as the point where W is largest in each iteration, and this is recorded as a pre-selected inflection point. Until the k-th iteration, when k×N is greater than the number of feature points in A1', the feature point at the beginning of the light stripe is used as the endpoint for calculating the pre-selected inflection point, and the iteration ends. The point with the largest W among all pre-selected inflection points is recorded as the inflection point. For example... Figure 2 As shown, taking A1' with 54 feature points and a preset iteration point number N of 20 as an example, the endpoint P in the first iteration... x 1 For distance P n Points spaced 20 feature points apart, take P n With P x 1 The feature point with the largest distance from the line connecting them is the first pre-selected inflection point; in the second iteration, the reference point P... x 2 For distance P n Points spaced 40 feature points apart, take P n With P x 2The feature point with the largest distance from the line connecting them is the second pre-selected inflection point; in the third iteration, since the number of iteration points 3×20=60 is greater than the number of feature points 54 in A1', the last feature point in A1', i.e., the starting feature point, is taken as the endpoint P along the direction of the light stripe extension. x 3 Take P n With P x 3 The feature point with the largest distance from the line connecting them is the third pre-selected inflection point; finally, by comparing the W values ​​of the three pre-selected inflection points, the one with the largest W value is taken as the inflection point.

[0050] Specifically, the method for extracting the center of the light stripe in step 1) is as follows: After projecting structured light onto the surface of the gap contour and taking a picture of the light stripe, select the ROI in the image, and extract the center point of the light stripe in the light stripe image within the ROI using the Steger algorithm or the gray-scale centroid method; wherein, the information of the center point of the light stripe includes, but is not limited to, coordinate values ​​and normal vectors; those skilled in the art may also choose other light stripe center extraction methods according to the purpose to be achieved.

[0051] The implementation of this method effectively prevents the intervention of subjective human factors. All data is processed using a unified evaluation standard, which effectively ensures the standardization and normalization of data processing and avoids human differences and errors. In addition, compared with the existing methods, this invention can trace the type identification data of each gap, which can support quality traceability and problem investigation, and effectively improve quality control capabilities.

[0052] A gap type recognition method based on machine vision, such as Figure 3 As shown, after obtaining the single-sided light stripe type of the gap using a machine vision-based gap single-sided light stripe type recognition method, the gap type is output according to the following rules:

[0053] The types of gaps include sharp edge gaps, beveled edge gaps, rolled edge gaps, and T-type gaps;

[0054] The sharp-edge gap consists of single-sided light strips on both sides of the gap, both of which are single-section light strips.

[0055] The oblique gap is a type of three-segment light strip with light strips on either side of the gap.

[0056] The rolled edge gap includes double rolled edge gap and sharp rolled edge gap; double rolled edge gap is when the light strips on both sides of the gap are of the two-segment light strip type; sharp rolled edge gap is when the light strips on one side of the gap are of the one-segment light strip type and the light strips on the other side are of the two-segment light strip type, and the supplementary judgment conditions are not met.

[0057] A T-shaped gap is defined as a gap where one side of the light bar is a single-segment light bar and the other side of the light bar is a two-segment light bar, and any supplementary judgment condition is met;

[0058] The supplementary judgment condition is:

[0059] Condition 1: Perform straight line fitting on the point cloud from the starting point to the inflection point in the point cloud set A1' corresponding to the two-segment light stripe to obtain the outer straight line of the light stripe. Draw a perpendicular line H from the end feature point to the outer straight line of the light stripe and obtain the positional relationship between the one-segment light stripe and the perpendicular line H. If the one-segment light stripe is divided into two parts by the perpendicular line, then it is determined that Condition 1 is satisfied; otherwise, it is not satisfied.

[0060] Condition 2: Perform straight line fitting on the point clouds from the starting feature point to the inflection point in each of the two single-sided light stripe point cloud sets A1', and calculate the angle between the two fitted straight lines. If the angle exceeds the preset angle threshold, then Condition 2 is satisfied; otherwise, Condition 2 is not satisfied. Preferably, the preset angle threshold is 30°~45°.

[0061] T-type gap is a special state of sharp rolled edge gap, and its subsequent gap evaluation method is different from that of conventional sharp rolled edge gap. Therefore, it is necessary to distinguish between sharp rolled edge gap and T-type gap. Thus, when the gap is identified as sharp rolled edge gap, it is necessary to simultaneously perform the judgment of supplementary judgment condition 1 and supplementary judgment condition 2. Only when both conditions are not met is the gap type determined to be rolled edge gap, otherwise it is T-type gap.

[0062] Furthermore, the types of beveled edge gaps include double beveled edge gaps, rolled beveled edge gaps, and sharp beveled edge gaps;

[0063] The double-bevel gap type is a three-segment light bar type with single-sided light bars on both sides.

[0064] The bevel gap type is one single-sided light bar type is three-segment light bar, and the other single-sided light bar is two-segment light bar;

[0065] The sharp bevel gap type is a gap with one single-sided light bar type being a three-segment light bar, and the other single-sided light bar being a one-segment light bar.

[0066] To more intuitively illustrate the correspondence between the two single-sided light stripe types and the gap type, the correspondence between the two single-sided light stripe types and the gap type is summarized as follows:

[0067]

[0068] In the table, "\" indicates that no judgment is required, "√" indicates that the condition is met or satisfied, and "×" indicates that the condition is not met or not satisfied.

[0069] The implementation of this method significantly improves the automation level of the gap visual measurement system, providing an efficient and convenient implementation path for gap size measurement. Furthermore, in practical applications, this method can be combined with gap measurement algorithms to achieve intelligent matching, breaking away from the inefficient mode of manually selecting gap types and matching algorithms in existing technologies. This not only reduces errors caused by manual intervention but also greatly improves the overall efficiency of measurement operations, providing strong support for the intelligent application of gap visual measurement technology.

[0070] For ease of interpretation and precise definition of the appended claims, the terms “upper,” “lower,” “left,” and “right” describe exemplary embodiments of the feature positions.

[0071] The foregoing description of specific exemplary embodiments of the present invention is for illustrative and descriptive purposes. It is not intended to be exhaustive, nor to limit the invention to the precise forms disclosed; obviously, many changes and variations are possible in accordance with the foregoing teachings. The exemplary embodiments were chosen and described to explain the specific principles of the invention and its practical application, thereby enabling others skilled in the art to implement and utilize various exemplary embodiments of the invention, as well as their different alternatives and modifications. The scope of the invention is intended to be defined by the appended claims and their equivalents.

Claims

1. A machine vision-based method for identifying the type of a single-sided light stripe in a gap, wherein the light stripe is perpendicular to the direction of gap extension during testing; characterized in that... Includes the following steps: 1) Extract the center of the light stripe from the image of the light stripe contour to obtain the center point set A of the light stripe; denote the position of the light stripe close to the gap as the end and the position far from the gap as the beginning; denote the center points of the light stripe arranged in an orderly manner between the beginning and end of the light stripe on one side as the point set A1; 2) Transform the data in point set A1 from two-dimensional coordinates in the image coordinate system to three-dimensional coordinates in the camera coordinate system, and then to two-dimensional coordinates in the light plane coordinate system, denoted as point cloud set A1'; 3) Process the data in the point cloud A1', select the feature points corresponding to the light stripe inflection point, the starting feature point and the ending feature point, and calculate the distance D from the light stripe inflection point to the straight line containing the starting feature point and the ending feature point based on them; Set a distance threshold ThD. If D < ThD, then the light stripe is a single-segment light stripe; otherwise, proceed to step 4). 4) Find the second inflection point between the inflection point and the end feature point. Denote the feature points between the inflection point and the second inflection point as the arc segment point set C1, and denote the feature points between the second inflection point and the end feature point as the straight line segment point set D1. Set a distance threshold ThL; perform circular arc fitting on the point cloud C1, and calculate the distance L1 from each feature point in C1 to the fitted circular arc; if L1 < ThL, then mark the feature point as an arc point. Perform a straight line fitting on the point cloud D1, and calculate the distance L2 from each feature point in D1 to the fitted straight line; if L2 < ThL, then mark the feature point as a straight line point. If the number of feature points marked as straight lines is greater than the number of feature points marked as arcs, and the number of feature points in point cloud D1 is greater than a preset threshold, then the light stripe is a three-segment light stripe; otherwise, it is a two-segment light stripe.

2. The method for identifying the type of single-sided light stripe in a gap based on machine vision as described in claim 1, characterized in that: Step 1) The method for obtaining the point set A1 is as follows: Starting from the starting endpoint P1 of the light stripe, in the light stripe center point set A, use the nearest neighbor search algorithm with a preset threshold to search for its nearest point P2 in the remaining light stripe center point set. Then remove P2 from the remaining light stripe center point set and use it as the new starting point. Search for the nearest point P3 of P2 until the search point P is reached. n The process stops when no nearest point is found within a preset threshold. Then, points P1, P2, P3, ..., P... n The set of center points of the light stripe is A1; P n That is, the end point of the light stripe.

3. The method for identifying the type of single-sided light stripe in a gap based on machine vision as described in claim 1, characterized in that: The method for selecting the inflection point of the light stripe in step 3) is as follows: taking the end feature point in the point cloud A1' as the starting point, perform iterative calculations towards the starting feature point of the light stripe. In the k-th iteration, select the feature point that is k×N points away from the starting point as the endpoint, where k is a positive integer and N is the preset number of points. In each iteration, calculate the distance W from all feature points between the starting point and the endpoint to the straight line between the starting point and the endpoint. Record the point with the largest W in each iteration as a pre-selected inflection point. When k×N is greater than the number of feature points in A1', the feature point at the beginning of the light stripe is used as the endpoint of this iteration to calculate the pre-selected inflection point and the iteration ends; the point with the largest W among all pre-selected inflection points is recorded as the inflection point.

4. The machine vision-based method for identifying the type of single-sided light stripe in a gap, as described in claim 1, is characterized in that: The method for extracting the center of the light stripe in step 1) is: Steger algorithm or gray-scale centroid method.

5. A gap type recognition method based on machine vision, which uses the method described in claim 1 to obtain the type of light stripes on both sides of the gap respectively, characterized in that: Gap types include sharp edge gap, beveled edge gap, rolled edge gap, and T-type gap; the gap type is determined according to the following rules: The sharp-edge gap is a single-segment light bar on both sides of the gap; The oblique gap is a three-segment light strip with light strips on either side of the gap; The rolled edge gap includes a double rolled edge gap and a sharp rolled edge gap; the double rolled edge gap is a gap in which both single-sided light stripes are of the two-segment light strip type; the sharp rolled edge gap is a gap in which one single-sided light strip is of the one-segment light strip type and the other single-sided light strip is of the two-segment light strip type, and the supplementary judgment conditions are not met. The T-shaped gap is defined as a single-sided light stripe of type one and a two-sided light stripe of type two, and satisfies any of the supplementary judgment conditions. The supplementary judgment condition is: Condition 1: Perform straight line fitting on the point cloud from the starting point to the inflection point in the point cloud set A1' corresponding to the two-segment light stripe to obtain the outer straight line of the light stripe. Draw a perpendicular line from the end feature point to the outer straight line of the light stripe to obtain the positional relationship between the one-segment light stripe and the perpendicular line. If the one-segment light stripe is divided into two parts by the perpendicular line, then it is determined that Condition 1 is satisfied; otherwise, it is not satisfied. Condition 2: Perform straight line fitting on the point cloud from the starting feature point to the inflection point in the two single-sided light stripe point cloud sets A1' respectively, and calculate the angle between the two fitted straight lines. If the angle exceeds the preset angle threshold, then condition 2 is satisfied; otherwise, condition 2 is not satisfied.

6. A gap type recognition method based on machine vision as described in claim 5, characterized in that: The types of beveled edge gaps include double beveled edge gaps, rolled beveled edge gaps, and sharp beveled edge gaps; The double-beveled gap type is that both single-sided light strips are three-segment light strips; The type of the beveled edge gap is that one single-sided light strip is a three-segment light strip, and the other single-sided light strip is a two-segment light strip; The sharp bevel gap type is that one single-sided light bar is a three-segment light bar, and the other single-sided light bar is a one-segment light bar.

Images