A method and system for visual inspection of component assembly standards

Abstract

The present disclosure provides a method and system for visual inspection of component assembly compliance, including: obtaining a first operation step and first operation video data that comply with the component assembly specification; determining a tracking target, an identification target and a first detection perspective based on the first operation step and the first operation video data, as well as a second operation step under each first detection perspective in the first operation step; collecting the second operation video data of component assembly in real time; using a target detection model to determine first motion trajectory data of the tracking target for the second operation video data corresponding to each second operation step under the first detection perspective; using the target detection model to determine a first detection area corresponding to the identification target; and determining whether the component assembly in each second operation step complies with the operation specification based on whether the first motion trajectory data of the tracking target is located within the first detection area corresponding to the identification target.

Description

Technical Field

[0001] The present disclosure relates to the field of industrial monitoring technology, and in particular to a method and system for visually inspecting component assembly standards. Background Art

[0002] In modern industrial production, parts assembly is a core step in the manufacturing process, and its quality directly impacts product performance and reliability. Despite advances in automation technology, in many production scenarios, parts assembly still relies heavily on manual labor. To ensure assembly quality, real-time monitoring and compliance testing of operators' operational processes are crucial. However, traditional testing methods face numerous challenges in practical application and fail to meet the demands of efficient and precise production. For example, testing operator compliance during parts assembly relies primarily on manual supervision and empirical judgment. For example, on-site quality management personnel visually inspect operators' movements to ensure they adhere to standard procedures, or conduct regular spot checks to assess assembly quality. However, this approach is susceptible to subjective factors such as fatigue, experience differences, and lack of concentration, leading to inconsistent and inaccurate test results. Different quality management personnel may have significant discrepancies in their judgment of the same operation, increasing quality control risks. Quality management requires significant time and manpower, especially in large-scale production environments, where real-time monitoring and comprehensive coverage of every operator is difficult to achieve. For example, in automotive parts assembly plants, the diverse variety of parts and complex assembly processes make it difficult for quality management personnel to complete a detailed assessment of all operations within a short period of time, limiting production efficiency. In addition, it may lead to delayed discovery of problems, which in turn affects production schedules. Summary of the Invention

[0003] The embodiments of the present disclosure provide a method and system for visual inspection of component assembly compliance, which is used to solve the problem that the existing component assembly process mainly relies on manual supervision for operator operation compliance inspection, which is difficult to meet the needs of efficient and accurate production.

[0004] In view of the above problems, in a first aspect, an embodiment of the present disclosure provides a method for visually inspecting component assembly standards, comprising:

[0005] Acquire first operation steps and first operation video data that comply with component assembly specifications;

[0006] Determining, based on the first operation step and the first operation video data, a tracking target, an identification target, and a first detection viewing angle, as well as a second operation step under each first detection viewing angle in the first operation step;

[0007] Real-time collection of second operation video data of parts assembly;

[0008] For the second operation video data corresponding to each second operation step under the first detection perspective, using the target detection model to determine the first motion trajectory data of the tracked target; and using the target detection model to determine the first detection area corresponding to the identified target;

[0009] Whether the assembly of parts in each second operation step complies with an operation specification is determined according to whether the first motion trajectory data of the tracked target is located in a first detection area corresponding to the identified target.

[0010] In a second aspect, a visual inspection system for component assembly compliance is provided, comprising:

[0011] A training module is configured to obtain a first operation step and first operation video data that conform to a component assembly specification; determine a tracking target, an identification target, and a first detection angle of view based on the first operation step and the first operation video data, as well as a second operation step at each first detection angle of view within the first operation step;

[0012] An acquisition module, for acquiring second operation video data of component assembly in real time;

[0013] The detection module is used to use the target detection model to determine the first motion trajectory data of the tracked target for the second operation video data corresponding to each second operation step under the first detection perspective; use the target detection model to determine the first detection area corresponding to the identified target; and determine whether the assembly of the parts in each second operation step complies with the operating specifications based on whether the first motion trajectory data of the tracked target is located within the first detection area corresponding to the identified target.

[0014] The beneficial effects of the embodiments of the present disclosure include:

[0015] The present disclosure provides a method and system for visual inspection of component assembly compliance, including: obtaining a first operation step and first operation video data that conform to component assembly compliance; determining a tracking target, an identification target, and a first detection perspective based on the first operation step and the first operation video data, as well as a second operation step under each first detection perspective in the first operation step; collecting second operation video data of component assembly in real time; using a target detection model to determine first motion trajectory data of the tracking target for the second operation video data corresponding to each second operation step under the first detection perspective; using the target detection model to determine a first detection area corresponding to the identification target; and determining whether the component assembly in each second operation step conforms to the operational specification based on whether the first motion trajectory data of the tracking target is within the first detection area corresponding to the identification target. The present disclosure provides a method for visual inspection of component assembly compliance, which monitors the first motion trajectory data of the target component and determines whether it is within the corresponding first detection area, thereby determining whether the component assembly in each second operation step conforms to the operational specification. The method can accurately identify behaviors that do not conform to operational specifications, effectively avoid component assembly quality problems caused by human negligence or improper operation, and significantly improve product yield. In addition, real-time detection and feedback of the parts assembly process can be achieved, which greatly reduces the time cost required for manual supervision, prevents rework and production delays caused by unqualified products flowing into the next process, and thus improves the operating efficiency of the entire production line. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A flowchart of a method for visual inspection of component assembly standards provided in an embodiment of the present disclosure;

[0017] Figure 2 A schematic diagram of tracking and identifying targets provided by an embodiment of the present disclosure;

[0018] Figure 3 This is a structural diagram of the component assembly standardization visual inspection system provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0019] The present disclosure provides a method and system for visual inspection of component assembly compliance. Preferred embodiments of the present disclosure are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are intended only to illustrate and explain the present disclosure and are not intended to limit the present disclosure. Furthermore, the embodiments and features within the embodiments of the present disclosure may be combined with one another unless there is a conflict.

[0020] The present disclosure provides a method and system for visual inspection of component assembly standards. Figure 1 Shown, including:

[0021] S101, obtaining a first operation step and first operation video data that conform to a component assembly specification;

[0022] S102, determining a tracking target, an identification target, and a first detection viewing angle, as well as a second operation step at each first detection viewing angle in the first operation step, based on the first operation step and the first operation video data;

[0023] S103, collecting second operation video data of component assembly in real time;

[0024] S104: for each second operation step corresponding to the second operation video data in the first detection perspective, using the target detection model to determine first motion trajectory data of the tracking target; and using the target detection model to determine a first detection area corresponding to the identified target;

[0025] S105 , determining whether the assembly of parts in each second operation step complies with an operation specification based on whether the first motion trajectory data of the tracked target is located within a first detection area corresponding to the identified target.

[0026] In modern industrial production, component assembly is a core step in the manufacturing process, and its quality directly impacts product performance and reliability. Whether it's precision electronics or complex automotive manufacturing, the accuracy and standardization of component assembly are crucial factors in ensuring final product quality. However, despite continuous advancements in automation technology, component assembly in many production scenarios still relies heavily on manual labor. This reliance presents numerous challenges, particularly in the area of ​​operational standardization. Traditional inspection methods are no longer able to meet today's demands for efficient and precise production. Taking automotive manufacturing as an example, the inspection of operator operational standardization during component assembly relies primarily on manual supervision and empirical judgment. On-site quality management personnel visually inspect operators' movements to ensure they adhere to standard procedures, or conduct regular spot checks to assess assembly quality. However, this approach presents numerous challenges. First, manual supervision is susceptible to subjective factors. Fatigue, experience differences, and inattention can lead to inconsistent and inaccurate monitoring results. For example, an experienced quality management personnel may be able to detect subtle operator errors, while a less experienced one may overlook them. Different quality management personnel can also have significant discrepancies in their judgment of the same operation, increasing quality control risks. Secondly, manual supervision is inefficient. Quality management personnel need to spend a lot of time and manpower, especially in large-scale production environments, and it is difficult to achieve real-time supervision and comprehensive coverage of every operator. Take automobile parts assembly plants as an example. The wide variety of parts and complex assembly processes make it difficult for quality management personnel to complete a detailed evaluation of all operations in a short period of time. This not only limits production efficiency, but may also lead to delayed discovery of problems, thereby affecting production progress. For example, on a busy automobile assembly line, hundreds of parts need to be assembled precisely, and manual supervision may not be able to detect assembly errors in a particular part in time. The problem may not be discovered until the product enters the subsequent process or even the final inspection stage. This not only increases rework costs but may also cause delays in the entire production plan. Faced with these challenges, traditional inspection methods are clearly unable to meet the needs of modern industrial production.

[0027] In the embodiments of this disclosure, the automotive parts assembly process is used as an example. During automotive parts assembly, the first step is to define the assembly specifications. These specifications typically exist in the form of standard operating procedures (SOPs), which include the assembly sequence, position requirements, and tightening torque requirements for each component. For example, during engine assembly, crankshaft installation requires first installing the bearings in place, then slowly inserting and securing the crankshaft. These steps are recorded in detail, forming the first operation step. This first operation step can include data types such as text, images, audio, and video, and contains information such as the full-process inspection steps, operating instructions, quality standards, inspection methods, and trajectory diagrams. A high-definition camera is used to capture the operation process that complies with the specifications, generating first operation video data. The standardized first operation video data must be consistent with the actual work scene, and the shooting angle must be maximized and unobstructed. During automotive parts assembly, the assembly of certain key components requires special attention. For example, in automobile body assembly, door installation is a critical step. Multiple first inspection angles can be used, each monitoring whether the corresponding second operation step complies with the specifications. The first inspection angle is the optimal camera angle to ensure that key information during the assembly process is clearly captured. For example, a camera can be installed on the side of a car door installation station to shoot the door installation process at a vertical angle. In addition, the first operation step is divided into more detailed second operation steps, such as positioning the car door, tightening the bolts, etc., and each step corresponds to a first detection angle of view. Based on the first operation video data, the operation steps performed at the same detection angle of view are observed to determine the number of angles of view of the first detection angle of view and the second operation steps at each first detection angle of view. The second operation step is the operation step that is divided from the first operation step and belongs to the same first detection angle of view. Figure 2 As shown, under the first detection angle of view, the tracking target 201 and the identification target 202 are determined. The tracking target 201 can be a tool or hand used by the operator, and usually runs through the entire inspection process. The identification target 202 can be a target that appears specifically in any first operation step or a target that appears together in multiple first operation steps, such as a car door and the installation position of the car door. In the actual assembly process, the high-definition camera installed on the production line is used to collect the operation video of the assembly worker in real time as the second operation video data. These cameras can be fixed or movable and are arranged according to the requirements of the first detection angle of view. For example, on an automobile engine assembly line, a camera can be installed above the engine assembly station to capture the process of workers installing pistons, crankshafts and other components in real time.

[0028] Furthermore, for each second operation step under the first detection perspective, the target detection model is used to detect the motion trajectory of the tracking target 201 in the second operation video data in real time in the order of the second operation steps to obtain first motion trajectory data of the tracking target 201; and the target detection model is used to determine the first detection area corresponding to the identification target 202. For example, the tracking target 201 is a vehicle door installation tool, and the identification target 202 is the vehicle door and its installation location. During the vehicle door installation process, the target detection model can track the motion trajectory of the vehicle door installation tool to obtain first motion trajectory data, and the target detection model can identify the vehicle door and its installation location to determine the first detection area. Based on whether the first motion trajectory data of the tracking target 201 is within the first detection area corresponding to the identification target 202, it is determined whether the assembly of parts in each second operation step complies with the operating specifications. For example, during the door installation process, if the first motion trajectory data of the door installation tool is within the first detection area corresponding to the door and the door installation position (such as the door installation hole), and the order and position of the door installation tool in the first motion trajectory data comply with the second operation step, then it is judged that the component assembly operation complies with the operating specifications; otherwise, an alarm will be issued and relevant information will be recorded.

[0029] In this embodiment of the present application, whether the assembly of parts in each second operation step complies with the operating specifications is determined by determining whether the first motion trajectory data of the tracked target is within the first detection area corresponding to the identified target. This can accurately identify non-standard operating behaviors, avoid quality issues in part assembly caused by human negligence or improper operation, and significantly improve product yield. This enables real-time detection and feedback, reduces the time cost of manual supervision, avoids rework and delays caused by unqualified products entering the next process, and improves the operating efficiency of the production line.

[0030] In another embodiment of the present disclosure, in step S104, determining the first detection area corresponding to the identified target using the target detection model includes:

[0031] The target detection model is used to determine a second detection area of ​​a direct positioning type corresponding to the identified target, including:

[0032] Step 1: Use the target detection model to determine the vertex information set of the detection box corresponding to the identified target, expressed as: , for each The coordinates are expressed as: , i ∈{1, 2, 3, 4};

[0033] Step 2: Compare each vertex separately coordinate The value of the size and The value of is used to determine the vertex coordinates from the vertex information set;

[0034] Step 3: Determine a second detection area of ​​a direct positioning type corresponding to the recognition target according to the vertex coordinates.

[0035] In the disclosed embodiment, the vertex information of the target detection frame is obtained, and the vertex coordinates are determined by coordinate comparison, thereby determining the second detection area of ​​the direct positioning type. Regarding step 1 above, in an automotive parts assembly scenario, for example, detecting the installation position of an automotive engine valve, the valve is used as the recognition target, and the target detection model is used to process the second operation video data including the valve. The target detection model generates a detection frame for the valve and determines the information of the four vertex corners of the detection frame, which is integrated into a vertex angle information set, represented as: , for each vertex in the set, its position in the image plane is represented by a two-dimensional coordinate, each The coordinates are expressed as: , i∈{1, 2, 3, 4}. For the above step 2, compare each vertex coordinate The value of the size and For example, determine the lower right corner vertex , from the set Select The two points with the largest coordinates are denoted as and ,exist and , select The point with larger coordinates is ; Determine the upper right vertex , from the set Remove the dot , select from the remaining three points The point with the largest coordinates is ; Determine the upper left vertex , from the set Remove the dot and , choose the remaining two points The point with smaller coordinates is , the remaining point is the lower right vertex Then, vertex coordinates are determined from the vertex angle information set. For step 3 above, a second detection area of ​​the direct positioning type corresponding to the identified target is determined based on the vertex coordinates. A rectangle can be constructed based on the vertex coordinates, and this rectangular area serves as the second detection area of ​​the direct positioning type corresponding to the identified target. Using the determined vertex coordinates as the boundary, a rectangular area is delineated in the second operation video data that accurately encompasses the valve location. Subsequent judgments can be made regarding the second detection area, such as whether the relative position of the valve and surrounding components complies with assembly specifications, or whether operations such as detecting surface defects on the valve comply with operating specifications. Automated acquisition and processing of vertex angle information based on the target detection model reduces the time cost of manual intervention and manual area delineation, enabling rapid location of the detection area. In large-scale automotive parts assembly inspection scenarios, this can significantly improve inspection efficiency and meet the requirements of real-time inspection on production lines. The second detection area of ​​the direct positioning type does not rely on specific part shapes or image features. As long as the target detection model can accurately generate a detection frame, the second detection area can be determined by processing vertex angle information. This makes it suitable for inspecting a wide range of automotive parts, demonstrating strong versatility and adaptability.

[0036] In another embodiment of the present disclosure, in step S104, determining the first detection area corresponding to the identified target using the target detection model includes:

[0037] The target detection model is used to determine a third detection area of ​​the angular positioning type corresponding to the identified target, including:

[0038] Step 1: Using any vertex coordinate of the second detection area as the reference vertex coordinate, scale the side length of the second detection area according to the scaling factor to determine the scaled vertex coordinate. The formula is:

[0039]

[0040] in, Represents the coordinates of the base vertex; 、 Indicates the coordinates of two vertices adjacent to the reference vertex coordinates in the second detection area; 、 represents the scaling factor; 、 、 、 Represents the scaled vertex coordinates; 、 Vectors representing the coordinates of two vertices adjacent to the reference vertex coordinate in the second detection area to the reference vertex coordinate;

[0041] Step 2: Determine a third detection area of ​​the angular positioning type corresponding to the recognition target according to the scaled vertex coordinates.

[0042] In the disclosed embodiment, based on the determined second detection area, any vertex is selected as a reference, the side length of the second detection area is scaled using a scaling factor, and the vertex coordinates are recalculated to determine the third detection area of ​​the angular positioning type, thereby achieving a more accurate or targeted detection range definition for the identified target. Regarding step 1 above, the third detection area of ​​the angular positioning type needs to use the coordinates of any vertex in the second detection area as the reference vertex coordinates, and is obtained by translating the sub-segments or extended segments of the two sides of the reference vertex coordinates. The formula is as follows:

[0043]

[0044] in, Represents the coordinates of the base vertex; 、 Indicates the coordinates of two vertices adjacent to the reference vertex coordinates in the second detection area; 、 represents the scaling factor; 、 、 、 Represents the scaled vertex coordinates. Exemplarily, whether the assembly of parts in the second operation step complies with the operating specifications needs to be determined based on the lower right corner of the second detection area. The lower right corner vertex coordinates of the second detection area are used as the reference vertex coordinates to subdivide the second detection area to obtain the third detection area. The lower side of the third detection area is equal to the lower side of the second detection area, and the right side of the third detection area is half of the right side of the third detection area. The lower left vertex coordinates and the upper right vertex coordinates of the third detection area are obtained, and the upper left vertex coordinates are obtained by the intersection of the lower side of the second detection area and the right side of the second detection area moving in parallel. That is, in the formula: is the lower side vector of the second detection area, is the right side vector of the second detection area, is the horizontal scaling factor, the value is 1, is the vertical scaling factor, the value is 0.5, 、 、 、 is the vertex coordinate of the third detection area. The other vertex coordinates of the second detection area are used as the reference vertex coordinates to obtain the vertex coordinates of the third detection area. The principle is the same as that of obtaining the vertex coordinates of the third detection area through the lower right corner, which will not be repeated here. For the above step 2, according to the scaled vertex coordinates, a rectangular area is delineated in the second operation video data as the third detection area. By setting different scaling factors, the second detection area can be flexibly expanded or reduced according to actual needs. In the detection of component assembly, key connection parts, fragile corners and other areas of the components can be focused on to improve the pertinence and effectiveness of detecting whether the assembly of components complies with the operating specifications, and avoid incomplete detection or misjudgment problems caused by excessively large or small detection ranges. When there are other components or background interference around the identification target, the third detection area can be determined by scaling the second detection area, which can effectively eliminate unnecessary interference information, make the detection more focused on the identification target itself, and improve the accuracy and reliability of the detection. It does not rely on specific target shapes and sizes. Regardless of the size of the identification target or the assembly environment it is in, the appropriate third detection area can be determined by adjusting the reference vertex coordinates and scaling factors. It has strong versatility and is suitable for assembly inspection scenarios of different types of products, providing a more flexible solution for automated inspection.

[0045] In another embodiment of the present disclosure, in step S104, determining the first detection area corresponding to the identified target using the target detection model includes:

[0046] The target detection model is used to determine a fourth detection area of ​​the angular rotation positioning type corresponding to the identified target, including:

[0047] Step 1: Determine the rotation angle ;

[0048] Step 2: According to the rotation angle , determine the rotation arc , the formula is:

[0049] ;

[0050] Step 3: According to the rotation arc , determine the rotation matrix , the formula is:

[0051] ;

[0052] Step 4: Using any scaled vertex coordinate of the third detection area as the rotation center coordinate, determine the translation vector of the scaled vertex coordinate of the third detection area. The formula is expressed as:

[0053]

[0054] in, represents the coordinates of the rotation center, Represents the scaled vertex coordinates, The translation vector representing the scaled vertex coordinates;

[0055] Step 5: According to the rotation matrix and translation vectors , determine the coordinates of the vertex after rotation , the formula is:

[0056]

[0057]

[0058] Step 6: Determine a fourth detection area of ​​the angular rotation positioning type corresponding to the identification target based on the rotated vertex coordinates.

[0059] In the embodiment of the present disclosure, a target detection model is used to determine the rotation angle, which is converted into a rotation radian, and then a rotation matrix is ​​constructed; the vertex coordinates after scaling of the third detection area are used as the rotation center, and the translation vector is calculated; the vertex coordinates after rotation are determined by combining the rotation matrix and the translation vector, and finally the fourth detection area of ​​the angular rotation positioning type is defined to achieve accurate detection range definition of the identified target under the rotation angle. With respect to the above step 1, in order to further accurately identify the detection area corresponding to the target, the fourth detection area of ​​the angular rotation positioning type can be based on the third detection area of ​​the angular positioning type, or based on the second detection area of ​​the direct positioning type, with any vertex coordinate as the rotation center, and obtained through rotation transformation, and the rotation angle is For example, rotate 10 degrees counterclockwise. For step 2 above, according to the formula:

[0060]

[0061] The rotation angle Convert to rotation radians ;

[0062] For step 3 above, according to the rotation radian , using the formula:

[0063]

[0064] Determine the rotation matrix The rotation matrix includes the direction and angle of rotation, and can rotate points on the plane according to the specified angle. For the above step 4, the coordinates of any scaled vertex in the third detection area are used as the rotation center coordinates to determine the translation vector of the scaled vertex coordinates in the third detection area. The formula is expressed as:

[0065]

[0066] For example, using the coordinates of the lower right corner vertex of the third detection area as the rotation center coordinates, determine the translation vector of the upper right corner coordinate relative to the lower right corner coordinates. The translation vector is used to adjust the positional relationship of the vertex coordinates before and after the rotation to ensure the accuracy of the rotation operation. For step 5 above, apply the rotation matrix to the translation vector, which is expressed as:

[0067]

[0068] Then calculate the coordinates of the rotated vertex , the formula is:

[0069]

[0070] Therefore, the vertex coordinates of the fourth detection area of ​​the angular rotation positioning type are expressed as:

[0071]

[0072] By using an object detection model to determine the rotation angle and construct the corresponding rotation matrix and translation vector, precise inspection area delineation for the identified target from different angles is possible. During the standardized monitoring of complex component assembly operations, the assembly status and surface quality of the components can be comprehensively observed, avoiding blind spots caused by a single inspection angle and improving the comprehensiveness and accuracy of inspections. In actual production, many components have diverse assembly positions and orientations. Flexible adjustment of the rotation angle of the inspection area based on actual needs can adapt to various complex assembly scenarios. For example, in the assembly of precision instruments, even if components are installed at irregular angles, angular rotation positioning can be used to determine the appropriate inspection area, ensuring effective monitoring. Furthermore, the use of an object detection model based on mathematical models and formulas enables the close integration of computer vision and automated inspection systems. On industrial production lines, automatic rotation adjustment of the inspection area can be achieved, reducing manual intervention, improving inspection efficiency and automation, and providing strong technical support for intelligent manufacturing.

[0073] In another embodiment of the present disclosure, in step S104, determining first motion trajectory data of the tracked target using a target detection model includes:

[0074] Step 1: Use the target detection model to detect the tracking target and determine the first range of the tracking target;

[0075] Step 2: Use any point within the first range as the first motion trajectory data of the tracking target.

[0076] In the embodiment of the present disclosure, the first range of the tracking target is determined by the target detection model, and then any point within the range is selected as its first motion trajectory data, so as to realize the recording and tracking of the motion path of the tracking target. For the above step 1, the target detection model is used to detect the tracking target that appears in real time in the second operation video data, and the target detection model selects a detection frame including the tracking target in the second operation video data. In order to facilitate tracking the motion trajectory of the tracking target, the detection frame of the tracking target can be a rectangle. The first range of the tracking target can be the detection frame of the tracking target. For the above step 2, the lower left corner coordinates of the first range are used. and the upper right corner coordinates Determine the coordinates of the trajectory point of the tracking target. Based on the coordinate range of the first range, any point within the first range can be determined as the motion trajectory point of the tracking target. For example, when determining whether the assembly of parts complies with the operating specifications, most of the area of ​​the tracking target is needed. The center point of the first range can be used as the motion trajectory point of the tracking target. The formula for the coordinates of the motion trajectory point is expressed as:

[0077]

[0078] Compared to tracking the motion trajectory of every point on the tracking target, determining the first range and selecting any point as trajectory data greatly simplifies the calculation process, reduces the performance requirements for hardware equipment, and improves data processing efficiency while ensuring the validity of trajectory recording, which can better meet the needs of real-time detection on the production line. In the component assembly environment, the tracking target may experience slight jitter, surface reflections, etc., resulting in slight deviations in target edge recognition. Recording trajectory data based on the first range can effectively filter out these local interferences. As long as the tracking target as a whole is within the first range, the selected trajectory points can stably reflect the movement trend of the tracking target, improving the stability and reliability of the trajectory data. Regardless of whether the tracking target is a regular shape (such as a rectangle) or an irregular shape (such as a concave polygon), determining its first range and extracting the first motion trajectory data through the target detection model is not limited by the specific shape of the tracking target. It has wide applicability and can be applied to tracking and detection scenarios of various component assemblies.

[0079] In another embodiment of the present disclosure, in the above step S105, determining whether the assembly of components in each second operation step complies with the operation specification based on whether the first motion trajectory data of the tracked target is located within the first detection area corresponding to the identified target includes:

[0080] Step 1: for each second operation step corresponding to the second operation video data in the first detection perspective, determine the second frame number in which the first motion trajectory data of the tracked target is located within the first detection area corresponding to the identified target;

[0081] Step 2: When the ratio of the second frame number to the first frame number is greater than or equal to a first threshold, determining that the component assembly corresponding to the second operation step meets the operation specification.

[0082] In the embodiment of the present disclosure, the proportion of the number of frames in which the first motion trajectory data of the tracking target is located within the first detection area of ​​the identification target is compared with the set first threshold value to determine whether the assembly of parts complies with the operating specifications, providing a quantitative basis for assembly quality inspection. For the above step 1, for the second operation video data corresponding to each second operation step under the first detection perspective, it is determined frame by frame whether the first motion trajectory data of the tracking target is located within the first detection area corresponding to the identification target in the second operation video data of the first frame number. If the first motion trajectory data of the tracking target is located within the first detection area corresponding to the identification target, the number of frames is counted to obtain the second number of frames. The first motion trajectory data of the tracking target in the second operation video data with the second number of frames is located within the first detection area corresponding to the identification target. For the above step 2, if the ratio of the second number of frames to the first number of frames is greater than or equal to the first threshold value, the assembly of parts corresponding to the second operation step complies with the operating specifications. If the ratio of the second number of frames to the first number of frames is less than the first threshold value, the assembly of parts corresponding to the second operation step does not comply with the operating specifications. Compared to traditional manual subjective judgment based on experience, by comparing the frame ratio with the threshold, the judgment of assembly compliance is converted into a quantifiable indicator, reducing interference from human factors, making the judgment results more objective and consistent, and improving the fairness and accuracy of the detection. By separately detecting each second operation step, the specific link in the assembly process where the problem occurs can be accurately located. For example, in the door installation step, if the ratio does not meet the standard, the deviation of the step can be directly identified, which facilitates the operator to adjust the operation in a targeted manner and improve rework efficiency. In large-scale component assembly production, massive amounts of video data can be quickly processed to achieve efficient detection of each assembly step, meet the real-time quality monitoring needs of mass production, and ensure the stability of the overall product assembly quality.

[0083] In another embodiment of the present disclosure, further comprising:

[0084] Step 1: When the operation status of the next second operation step is not displayed and whether the component assembly of the current second operation step complies with the operation specifications has been determined, the operation status of the current second operation step is displayed; wherein the operation status includes: complies with the operation specifications, does not comply with the operation specifications, and is not displayed.

[0085] In the disclosed embodiment, whether the assembly of parts in the current second operation step complies with the operating specifications is presented in an orderly manner and timely feedback is given. Taking the assembly of automobile seat slide rails as an example, the entire assembly process includes multiple second operation steps, such as slide rail positioning, bolt pre-tightening, bolt tightening, etc. Assume that the current second operation step is "bolt pre-tightening", the next second operation step is "bolt tightening" and its operation status is not displayed. After completing the assembly specification inspection of the "bolt pre-tightening" step, if the inspection result is in compliance with the operating specifications, the current step status is immediately displayed as "compliant with the operating specifications" or a related symbol, such as a "check mark"; if the inspection result is not in compliance with the operating specifications, "not in compliance with the operating specifications" or a "cross" is displayed. For example, the second operation step under the first detection perspective includes steps 2 and 5, then the following situations exist:

[0086] , ,

[0087] , , ,

[0088] Where: is the second frame number corresponding to step 2, is the second frame number corresponding to step 5, Indicates the first frame number, represents the first threshold corresponding to step 2, represents the first threshold corresponding to step 5, Indicates whether the operation status of step 2 is displayed. Indicates whether the operation status of step 3 is displayed. Indicates whether the operation status of step 5 is displayed. Indicates whether the operation status of step 6 is displayed. Indicates the sequence number of the current second operation step, Indicates that it has been displayed, showing "Complies with operating specifications" or "Does not comply with operating specifications". Indicates not displayed.

[0089] This process ensures that the inspection results of the current second operation step are promptly presented to operators and management personnel even when the status of subsequent steps is not being performed. This ensures that the second operation step is carried out in an orderly and sequential manner. By promptly displaying the operation status of the current step, operators can immediately determine whether their operations comply with specifications. If not, they can immediately stop the machine for inspection and make corrections, avoiding the problem from being carried over to the next step. This reduces batch rework caused by operational errors and improves assembly efficiency. For example, during the seat rail bolt pre-tightening step, when the operator sees the prompt "Not in compliance with operating specifications," they can immediately adjust the pre-tightening tool parameters and re-operate to avoid affecting subsequent tightening steps due to improper bolt pre-tightening. Prioritizing the display of the current step status when the status of the next step is not yet displayed prevents information transmission from being interrupted while waiting for information from subsequent steps. It also allows production management personnel to clearly understand the assembly progress and quality status, facilitating real-time control of the production process and ensuring the orderly progress of the entire assembly process. The display of the operation status records the quality of each step. When assembly problems are discovered later, the specific link and responsible person of the problem can be quickly located by tracing back the display status of each step, providing a clear basis for quality traceability. At the same time, it can also encourage operators to pay more attention to operational standardization and improve overall assembly quality awareness.

[0090] In another embodiment of the present disclosure, in the above step 1, if the operation status of the next second operation step is not displayed and it has been determined whether the component assembly of the current second operation step complies with the operation specification, the operation status of the current second operation step is displayed, including:

[0091] Step 1: In the second operation video data corresponding to the current second operation step, if a tracking target is detected in the first three frames and the operation status of the next second operation step is not displayed, the operation status of the current second operation step is displayed based on whether the assembly of the parts corresponding to the current second operation step complies with the operation specification;

[0092] The method also includes:

[0093] Step 2: In the second operation video data corresponding to the current second operation step, if the tracking target is detected in the current frame and the tracking target is not detected in the previous third frames, obtain the operation status of all second operation steps;

[0094] Step 3: If the operation status of all the second operation steps is in compliance with the operation specifications, a voice broadcast check is performed to check if it is qualified;

[0095] Step 4: When the operation statuses of all second operation steps are in compliance with the operation specifications and in non-compliance with the operation specifications, the second operation steps that do not comply with the operation specifications are announced by voice;

[0096] Step 5: When the operation status of all second operation steps does not comply with the operation specifications, no voice broadcast is performed.

[0097] In the disclosed embodiment, the first three frames of the second operation video data are used to determine whether the current operation process has ended. If so, a voice announcement is made. Regarding step 1, if a tracking target is detected within the first three frames of the second operation video data corresponding to the current second operation step, this indicates that the operator is performing component assembly. If the operation status of the next second operation step is not displayed, the operation status of the current second operation step is displayed based on whether the component assembly of the current second operation step complies with the operating specifications. Regarding step 2, if a tracking target is detected in the current frame of the second operation video data corresponding to the current second operation step, and no tracking target is detected within the first three frames, this indicates that the current operation process has ended, and a voice announcement is made based on the operation status of all second operation steps. Regarding step 3, if the operation status of all second operation steps obtained complies with the operating specifications, a voice announcement is made stating that "Inspection Passed" to inform the operator that the assembly quality has met the standards. Regarding step 4, if all second steps both comply with the operating specifications and do not comply with them, the non-compliant steps are announced, facilitating quick correction by the operator. For step 5 above, if the operating status of all second operating steps does not meet the specifications, no voice broadcast will be made to avoid interference from invalid information. The operator can use the operating status information displayed on the screen to comprehensively check the problem. After the voice broadcast, the operating status of all second operating steps can be set to not be displayed, and the next round of operation process can be started. By detecting and tracking the target of the second operation video data of the first three frames, it is possible to quickly determine whether this round of operation process is completed. Differentiated voice broadcasts are made according to the operating status of all steps, affirmation is given when qualified, problems are pointed out accurately when there are problems, and no broadcast is made when all are unqualified, which not only improves communication efficiency but also avoids information redundancy. Timely status display and targeted voice broadcasts allow operators to quickly discover and solve problems, reduce the continuation of errors, thereby improving assembly quality and overall production efficiency, and are especially suitable for assembly scenarios of mass production.

[0098] In another embodiment of the present disclosure, the target detection model is trained in the following manner:

[0099] Step 1: labeling the first operation video data according to the tracking target, the identification target and the first detection perspective;

[0100] Step 2: Divide the labeled first operation video data into a training set and a validation set, and train the target detection model.

[0101] In the embodiment of the present disclosure, the first operation video data is targetedly labeled, and the labeled data is used as a training set to train the target detection model, so that the model can better adapt to the detection needs of tracking targets and identifying targets in actual detection scenarios. For the above step 1, since the first detection perspective of the detection scene is rotatable and movable, the target detection model needs to select a model that supports rotation detection. The selectable models include but are not limited to the obb model improved by YOLO, the Oriented R-CNN model, the RetinaNet improved model, etc. The first operation video data is marked at frame intervals. Uniformly decode the video frames. Select a labeling tool based on the target detection model and label the decoded frames according to the tracking targets and identification targets under different first detection perspectives. To facilitate tracking, tracking indicators can be labeled using horizontal rectangles, while there are no restrictions on labeling identification targets. To detect tracking indicators, the target detection model may include tracking algorithms such as BYTETracker, DeepSORT, and TrackFormer. Regarding step 2 above, the labeled first operation video data is proportionally divided into a training set and a validation set. The training set is then input into the target detection model to be trained. During training, the target detection model continuously learns the characteristics of the labeled tracking targets and identification targets, such as shape, size, and appearance under the first detection perspective. By repeatedly adjusting the model parameters, the target detection model accurately identifies the tracking targets and identification targets in the video frames and outputs their location information. During training, the model performance is evaluated in real time using the validation set. Training is terminated when the target detection model's detection accuracy on the validation set reaches a preset standard, resulting in a target detection model suitable for component assembly inspection. By labeling the tracking targets, identification targets, and first detection perspectives according to the actual detection scene, the training data is highly matched to the actual detection needs. The model can more accurately learn the characteristics of the target, thereby improving the detection accuracy of the target and reducing false detections and missed detections. Training for specific detection perspectives and targets allows the target detection model to better adapt to factors such as lighting and background interference in the scene, maintaining stable detection performance in actual applications without the need to frequently adjust model parameters to adapt to different scenarios. Labeling focuses on the targets and perspectives that need to be detected, avoiding interference from irrelevant information, enabling the target detection model to converge more quickly during training, shortening training time, while reducing the consumption of computing resources and lowering model training costs.

[0102] Based on the same disclosed concept, the embodiment of the present disclosure also provides a component assembly standardization visual inspection system. Since the principle of the problem solved by this system is similar to the aforementioned component assembly standardization visual inspection method, the implementation of this system can refer to the implementation of the aforementioned method, and the repeated parts will not be repeated.

[0103] The present disclosure provides a visual inspection system for component assembly standards. Figure 3 Shown, including:

[0104] Training module 301 is configured to obtain a first operation step and first operation video data that conform to a component assembly specification; determine a tracking target, an identification target, and a first detection angle of view based on the first operation step and the first operation video data, as well as a second operation step at each first detection angle of view within the first operation step;

[0105] The acquisition module 302 is used to acquire the second operation video data of component assembly in real time;

[0106] The detection module 303 is used to determine the first motion trajectory data of the tracked target using the target detection model for the second operation video data corresponding to each second operation step under the first detection perspective; determine the first detection area corresponding to the identified target using the target detection model; and determine whether the assembly of the parts in each second operation step complies with the operation specifications based on whether the first motion trajectory data of the tracked target is within the first detection area corresponding to the identified target.

[0107] In another embodiment of the present disclosure, the detection module 303 is configured to use a target detection model to determine a second detection area of ​​a direct positioning type corresponding to the identified target, including:

[0108] The target detection model is used to determine the vertex information set of the detection box corresponding to the identified target, which is expressed as: , for each The coordinates are expressed as: , i ∈{1, 2, 3, 4};

[0109] Compare each vertex separately coordinate The value of the size and Determine the vertex coordinates from the vertex angle information set according to the value of ;

[0110] A second detection area of ​​a direct positioning type corresponding to the identified target is determined according to the vertex coordinates.

[0111] In another embodiment of the present disclosure, the detection module 303 is further configured to use a target detection model to determine a third detection area of ​​the angular positioning type corresponding to the identified target, including:

[0112] Taking any vertex coordinate of the second detection area as the reference vertex coordinate, the side length of the second detection area is scaled according to the scaling factor to determine the scaled vertex coordinate, which is expressed as follows:

[0113]

[0114] in, Represents the coordinates of the base vertex; 、 Indicates the coordinates of two vertices adjacent to the reference vertex coordinates in the second detection area; 、 represents the scaling factor; 、 、 、 Represents the scaled vertex coordinates; 、 Vectors representing the coordinates of two vertices adjacent to the reference vertex coordinate in the second detection area to the reference vertex coordinate;

[0115] A third detection area of ​​the angular positioning type corresponding to the identification target is determined according to the scaled vertex coordinates.

[0116] In another embodiment of the present disclosure, the detection module 303 is further configured to use a target detection model to determine a fourth detection area of ​​the angular rotation positioning type corresponding to the identified target, including:

[0117] Determine the rotation angle ;

[0118] According to the rotation angle , determine the rotation arc , the formula is:

[0119] ;

[0120] According to the rotation arc , determine the rotation matrix , the formula is:

[0121] ;

[0122] Taking any scaled vertex coordinate of the third detection area as the rotation center coordinate, the translation vector of the scaled vertex coordinate of the third detection area is determined, and the formula is expressed as:

[0123]

[0124] in, represents the coordinates of the rotation center, Represents the scaled vertex coordinates, The translation vector representing the scaled vertex coordinates;

[0125] According to the rotation matrix and translation vectors , determine the coordinates of the vertex after rotation , the formula is:

[0126]

[0127]

[0128] A fourth detection area of ​​the angular rotation positioning type corresponding to the identification target is determined according to the rotated vertex coordinates.

[0129] In another embodiment of the present disclosure, the detection module 303 is configured to detect the tracked target using a target detection model to determine a first range where the tracked target is located;

[0130] Any point within the first range is used as first motion trajectory data of the tracking target.

[0131] In another embodiment of the present disclosure, the detection module 303 is configured to determine, for each second operation step corresponding to the second operation video data at the first detection perspective, a second frame number in which the first motion trajectory data of the tracked target is located within the first detection area corresponding to the identified target;

[0132] When the ratio of the second frame number to the first frame number is greater than or equal to a first threshold, it is determined that the component assembly corresponding to the second operation step complies with the operation specification.

[0133] In another embodiment of the present disclosure, the detection module 303 is also used to display the operation status of the current second operation step when the operation status of the next second operation step is not displayed and whether the assembly of parts of the current second operation step complies with the operation specifications has been determined; wherein the operation status includes: complying with the operation specifications, not complying with the operation specifications and not displayed.

[0134] In another embodiment of the present disclosure, the detection module 303 is further configured to, when a tracking target is detected in the first three frames of the second operation video data corresponding to the current second operation step and the operation status of the next second operation step is not displayed, display the operation status of the current second operation step based on whether the assembly of the parts corresponding to the current second operation step complies with the operation specification;

[0135] The detection module 303 is further configured to obtain the operation status of all second operation steps when, in the second operation video data corresponding to the current second operation step, the tracking target is detected in the current frame and the tracking target is not detected in the previous third frames;

[0136] If the operating status of all second operating steps is in compliance with the operating specifications, the voice broadcast inspection is performed to check if it is qualified;

[0137] In the case where the operation statuses of all second operation steps include those that comply with the operation specifications and those that do not comply with the operation specifications, a voice announcement is made for the second operation steps that do not comply with the operation specifications;

[0138] When the operation status of all second operation steps does not comply with the operation specifications, no voice announcement is made.

[0139] In another embodiment of the present disclosure, the training module 301 is configured to train the target detection model in the following manner:

[0140] labeling the first operation video data according to the tracking target, the identification target, and the first detection perspective;

[0141] The labeled first operation video data is divided into a training set and a validation set to train the object detection model.

[0142] Through the above description of the embodiments, those skilled in the art will clearly understand that the embodiments of the present disclosure can be implemented through hardware or through software plus the necessary general-purpose hardware platform. Based on this understanding, the technical solutions of the embodiments of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, or mobile hard drive) and includes a number of instructions for causing a computer device (such as a personal computer, server, or network device) to execute the methods described in the various embodiments of the present disclosure.

[0143] Those skilled in the art will understand that the accompanying drawings are merely schematic diagrams of a preferred embodiment, and the modules or processes in the accompanying drawings are not necessarily required for implementing the present disclosure.

[0144] Those skilled in the art will appreciate that the modules in the devices of the embodiments may be distributed in the devices of the embodiments as described in the embodiments, or may be located in one or more devices different from the embodiments with corresponding changes. The modules of the above embodiments may be combined into one module or further split into multiple submodules.

[0145] The serial numbers of the above-mentioned embodiments of the present disclosure are for description only and do not represent the advantages or disadvantages of the embodiments.

[0146] Obviously, those skilled in the art may make various changes and modifications to the present disclosure without departing from the spirit and scope of the present disclosure. Thus, if these modifications and variations of the present disclosure fall within the scope of the claims of the present disclosure and their equivalents, the present disclosure is intended to include these modifications and variations.

Claims

1. A method for visual inspection of component assembly standards, characterized in that: include: Acquire a first operation step and first operation video data that conform to component assembly specifications; Determining, based on the first operation step and the first operation video data, a tracking target, an identification target, and a first detection viewing angle, as well as a second operation step under each first detection viewing angle in the first operation step; Real-time collection of second operation video data of parts assembly; Determining first motion trajectory data of the tracked target using a target detection model for the second operation video data corresponding to each second operation step under the first detection perspective; Determining a first detection area corresponding to the identified target using a target detection model includes: By obtaining the vertex information of the target detection frame, the vertex coordinates are determined by coordinate comparison, and then the second detection area of ​​the direct positioning type is determined; Based on the determined second detection area, any vertex is selected as a reference, the side length of the second detection area is scaled using the scaling factor, and the vertex coordinates are recalculated to determine the third detection area of ​​the angular positioning type, thereby achieving a more accurate or targeted detection range definition for the identified target; The target detection model is used to determine the rotation angle, which is converted into rotation radians to construct a rotation matrix. The translation vector is calculated using the scaled vertex coordinates of the third detection area as the rotation center. The rotation matrix and translation vector are combined to determine the coordinates of the rotated vertex, and finally the fourth detection area of ​​the angular rotation positioning type is defined to achieve accurate detection range definition of the identified target under the rotation angle. Whether the assembly of parts in each second operation step complies with an operation specification is determined according to whether the first motion trajectory data of the tracked target is located in a first detection area corresponding to the identified target.

2. The method according to claim 1, wherein The adopting the target detection model to determine the first detection area corresponding to the identified target includes: Determining a second detection area of ​​a direct positioning type corresponding to the identified target using a target detection model includes: The target detection model is used to determine the vertex information set of the detection box corresponding to the identified target, which is expressed as: , for each The coordinates are expressed as: , i ∈{1, 2, 3, 4}; Compare each vertex separately coordinate The value of the size and Determine the vertex coordinates from the vertex angle information set according to the value of ; A second detection area of ​​a direct positioning type corresponding to the identified target is determined according to the vertex coordinates.

3. The method according to claim 2, wherein The determining of the first detection area corresponding to the identified target by using the target detection model further includes: Determining a third detection area of ​​the angular positioning type corresponding to the identified target using a target detection model includes: Taking any vertex coordinate of the second detection area as the reference vertex coordinate, the side length of the second detection area is scaled according to the scaling factor to determine the scaled vertex coordinate, which is expressed as follows: in, Represents the coordinates of the base vertex; 、 Indicates the coordinates of two vertices adjacent to the reference vertex coordinates in the second detection area; 、 represents the scaling factor; 、 、 、 Represents the scaled vertex coordinates; 、 Vectors representing the coordinates of two vertices adjacent to the reference vertex coordinate in the second detection area to the reference vertex coordinate; A third detection area of ​​the angular positioning type corresponding to the identification target is determined according to the scaled vertex coordinates.

4. The method according to claim 3, wherein The determining of the first detection area corresponding to the identified target by using the target detection model further includes: Determining a fourth detection area of ​​the angular rotation positioning type corresponding to the identified target using a target detection model includes: Determine the rotation angle ; According to the rotation angle , determine the rotation arc , the formula is: ； According to the rotation arc , determine the rotation matrix , the formula is: ； Taking any scaled vertex coordinate of the third detection area as the rotation center coordinate, the translation vector of the scaled vertex coordinate of the third detection area is determined, and the formula is expressed as: in, represents the coordinates of the rotation center, Represents the scaled vertex coordinates, The translation vector representing the scaled vertex coordinates; According to the rotation matrix and translation vectors , determine the coordinates of the vertex after rotation , the formula is: A fourth detection area of ​​the angular rotation positioning type corresponding to the identification target is determined according to the rotated vertex coordinates.

5. The method according to claim 1, wherein The determining of first motion trajectory data of the tracked target by using a target detection model includes: Detecting the tracked target using a target detection model to determine a first range where the tracked target is located; Any point within the first range is used as first motion trajectory data of the tracking target.

6. The method according to claim 1, wherein The determining whether the assembly of parts in each second operation step complies with the operation specification based on whether the first motion trajectory data of the tracked target is located in the first detection area corresponding to the identified target includes: For each second operation step corresponding to the second operation video data in the first detection perspective, determine a second frame number in which the first motion trajectory data of the tracked target is located within the first detection area corresponding to the identified target; When the ratio of the second frame number to the first frame number is greater than or equal to a first threshold, it is determined that the component assembly corresponding to the second operation step complies with the operation specification.

7. The method according to claim 1, wherein Also includes: When the operation status of the next second operation step is not displayed and whether the component assembly of the current second operation step complies with the operation specifications has been determined, the operation status of the current second operation step is displayed; wherein, the operation status includes: complying with the operation specifications, not complying with the operation specifications and not displayed.

8. The method according to claim 7, wherein When the operation status of the next second operation step is not displayed and it has been determined whether the assembly of parts in the current second operation step complies with the operation specification, displaying the operation status of the current second operation step includes: In the second operation video data corresponding to the current second operation step, if a tracking target is detected in the first three frames and the operation status of the next second operation step is not displayed, the operation status of the current second operation step is displayed based on whether the assembly of the parts corresponding to the current second operation step complies with the operation specification; The method further comprises: In the second operation video data corresponding to the current second operation step, if the tracking target is detected in the current frame and the tracking target is not detected in the previous third frames, obtaining the operation status of all second operation steps; If the operating status of all second operating steps is in compliance with the operating specifications, the voice broadcast inspection is carried out to check if it is qualified; In the case where the operation statuses of all second operation steps include those that comply with the operation specifications and those that do not comply with the operation specifications, a voice announcement is made for the second operation steps that do not comply with the operation specifications; When the operation status of all second operation steps does not comply with the operation specifications, no voice announcement is made.

9. The method according to claim 1, wherein The target detection model is trained in the following manner: labeling the first operation video data according to the tracking target, the identification target, and the first detection perspective; The labeled first operation video data is divided into a training set and a validation set to train the object detection model.

10. A visual inspection system for component assembly standardization, characterized in that: include: A training module, configured to obtain first operation steps and first operation video data that conform to component assembly specifications; Determining, based on the first operation step and the first operation video data, a tracking target, an identification target, and a first detection viewing angle, as well as a second operation step under each first detection viewing angle in the first operation step; An acquisition module, for acquiring second operation video data of component assembly in real time; a detection module, configured to determine, based on the second operation video data corresponding to each second operation step under the first detection perspective, first motion trajectory data of the tracked target using a target detection model; A target detection model is used to determine the first detection area corresponding to the identified target, including: obtaining the vertex angle information of the identified target detection frame, determining the vertex coordinates through coordinate comparison, and then determining the second detection area of ​​the direct positioning type; based on the determined second detection area, any vertex is selected as a reference, the side length of the second detection area is scaled by the scaling factor, and the vertex coordinates are recalculated to determine the third detection area of ​​the angular positioning type, thereby achieving a more accurate or targeted detection range definition for the identified target; the target detection model is used to determine the rotation angle, and the rotation angle is converted into a rotation radian, thereby constructing a rotation matrix; the translation vector is calculated with the scaled vertex coordinates of the third detection area as the rotation center; the rotated vertex coordinates are determined by combining the rotation matrix and the translation vector, and finally the fourth detection area of ​​the angular rotation positioning type is delineated to achieve accurate detection range definition for the identified target under the rotation angle; based on whether the first motion trajectory data of the tracked target is located in the first detection area corresponding to the identified target, it is determined whether the assembly of the components in each second operation step complies with the operating specifications.

Images