Method, device and system for identifying parts on a rotatably supported vehicle component

Abstract

The present invention relates to a method for identifying a part installed on a rotatably supported vehicle component, the method comprising: acquiring an image of the rotatably supported vehicle component; matching feature points of the acquired image based on a plurality of feature points preset in a template image; triangulating the acquired image based on the matched feature points; mapping each triangular region obtained by triangulation to a corresponding triangular region constructed by feature points preset in the template image by means of affine transformation to obtain a corrected image, thereby overcoming image distortion and allowing the region of the part to be restored; intercepting a predetermined region of the part from the corrected image; and identifying the part in the intercepted predetermined region. The present invention relates to a computer-readable storage medium, an apparatus for identifying a part installed on a rotatably supported vehicle component, a system for identifying a part installed on a rotatably supported vehicle component, and a vehicle production line.

Description

Technical Field

[0001] The present invention relates to a method for identifying a part mounted on a rotatably supported vehicle component, a computer-readable storage medium, an apparatus for identifying a part mounted on a rotatably supported vehicle component, a system for identifying a part mounted on a rotatably supported vehicle component, and a vehicle production line. Background Art

[0002] In the vehicle production line, although large-scale automated production has been achieved through robots and mechanical arms in stamping, body, painting and other workshops, a large amount of manual operation is still required, especially in the final assembly workshop for assembly, to ensure the process quality during the fine assembly process. In particular, in the final assembly workshop, workers will install the required parts on the rotatably supported vehicle parts, such as the dashboard, in sequence according to the work order corresponding to each vehicle part transported to the workstation. Each workstation has its own duties and is only responsible for installing the parts it is responsible for. For the convenience of installation, the vehicle parts may need to be rotated from 0 to 180 degrees as needed. As a result, the parts installed in a certain process are often covered by new parts in the next process or rotated to the visual blind spot.

[0003] However, the parts that need to be installed often play an important role in the quality of the vehicle, such as ensuring the positioning of various devices and equipment, preventing looseness, and avoiding water leakage. Here, take the clips installed under the dashboard as an example. The plastic seal fixed by these clips will play a vital role in the drainage of the air-conditioning pipes. If the clips are lost, the air-conditioning water will leak into the gaps in the car, causing corrosion of the lines and breeding of odors. Moreover, since the clip is close to the gear control wiring, the lack of the clip may even interfere with the sensitivity and stability of the gear control, causing safety hazards during driving. Such parts are often difficult to be checked again in subsequent processes. Moreover, during rework or repair, the center console must be disassembled as a whole before it can be reinstalled, and its repair cost is high and time-consuming.

[0004] Based on the above reasons, how to judge whether the required parts are correctly installed in the corresponding positions becomes the key to vehicle quality control.

[0005] Nowadays, for example, in the final assembly workshop of a vehicle, workers are required to use handheld mobile camera devices to take photos of the installed parts after manually installing the parts, so as to achieve the purpose of archiving, etc. However, such a process can only serve as a basis for post-inspection, and not only cannot correspond to the actual work order and thus does not have the function of real-time alarm, but also increases the workload of workers, which is not conducive to improving production efficiency. Summary of the invention

[0006] The task of the present invention is to provide a possibility of automatically identifying parts on a rotatably supported vehicle component in vehicle manufacturing, so as to realize real-time monitoring of the assembly process of the rotatably supported vehicle component, and quickly and accurately identify the parts to be installed with the lowest possible calculation amount, so as to ensure the installation of the parts on the vehicle component, avoid rework or repair due to missing or wrong installation, and improve the quality of the vehicle. Under the premise of not affecting the normal production of the original installation station, the manual operation of the production workers is reduced, the labor cost is reduced and the production efficiency is improved.

[0007] A first aspect of the invention relates to a method for identifying a part mounted on a rotatably supported vehicle component, the method comprising:

[0008] - acquiring an image of the rotatably supported vehicle component;

[0009] - Match the feature points of the acquired image based on multiple feature points preset in the template image;

[0010] - Triangulate the acquired image based on the matched feature points;

[0011] - Mapping each of the divided triangular regions to the corresponding triangular regions constructed by the preset feature points in the template image by means of affine transformation to obtain a corrected image, thereby overcoming the image distortion and restoring the region of the part;

[0012] - extracting a predetermined area of ​​the part from said corrected image;

[0013] - Identify parts within a predefined area captured.

[0014] In the present invention, rotatable support refers in particular to vehicle parts that can be supported rotatably around an axis, but may also include manual flipping of the position of the vehicle parts to change their position orientation, etc. However, such vehicle parts may appear at different angles in the acquired image due to adjustments made by workers, offsets of support points, dimensional errors of brackets, forces applied during installation, etc. In addition, positioning deviations on the guide rails and differences in specifications of vehicle parts, especially in mixed-line production or flexible production, may also occur in the production line. For these reasons, distortions may appear in the acquired two-dimensional images of vehicle parts, making it difficult to correct such images by simple translation, rotation or scaling, resulting in the inability to accurately locate in the image and difficulty in conventional template matching. Moreover, the size of the parts is usually much smaller than that of the vehicle parts as a whole, and conventional translation, rotation or scaling of the acquired vehicle parts images will inevitably cause local distortions, which seriously affect the accuracy of part recognition.

[0015] Based on the above background, the present invention provides a method for image recognition of parts installed on such vehicle parts. In the method, first, an image of a rotatably supported vehicle part on which the part should be installed is obtained. Next, feature point matching is performed. Here, a template image is preset for reference. Feature points are pre-extracted from the template image. Here, information used to distinguish images is expressed by extracting feature points. Feature points can be understood as positions with large changes in the image, which are unique, reflect the overall contour or change significantly. Feature point matching can be understood as searching for feature points in the acquired image, describing the feature points in the form of vectors, and finally achieving feature point matching by comparing the description vectors. In this way, points with the same or similar features as the template image can be obtained in the acquired image. Then, the acquired image is triangulated based on the matched feature points, especially Delaunay triangulation, so that the acquired image is divided into several triangular areas. Then, each of the divided triangular regions is mapped to the corresponding triangular regions constructed by the preset feature points in the template image by means of affine transformation, so as to obtain a corrected image, thereby overcoming the image distortion and allowing the region of the part to be restored. Here, each of the divided triangular regions is mapped or aligned to the corresponding triangular regions in the template image by means of affine transformation. In this way, localized deformation matching is achieved for the unique conditions of the image of the rotatably supported vehicle component, overcoming the image distortion caused by the worker's adjustment, the offset of the support point, the dimensional error of the bracket, the force applied during the installation process, and the like, and more accurately correcting to the state of the template image. Thus, the predetermined region of the part can be intercepted from the corrected image. Here, the predetermined region is the region where each part is marked in the template image. Finally, the part is identified in the intercepted predetermined region.

[0016] The present invention can automatically monitor the assembly process of rotatably supported vehicle parts in real time. Three-dimensional reconstruction of vehicle parts is not used in image processing, so that the image can be quickly and accurately corrected and the parts to be installed can be identified with the lowest possible amount of calculation. The part area can be restored by local deformation matching according to the present invention, ensuring the accuracy of identification. In addition, the correct installation of parts on vehicle parts can be ensured in time at the installation station, avoiding rework or repair due to missing or wrong installation, thereby improving the quality of the vehicle. Without affecting the normal production of the original installation station, the manual operation of production workers is reduced, the labor cost is reduced and the production efficiency is improved.

[0017] According to an embodiment of the present invention, the template image only involves the image region containing the parts. Here, only the image region containing the parts is used as the template image, so that a relatively small image region can be processed in the feature point matching, triangulation and mapping process, thereby reducing the amount of calculation, increasing the processing speed, and achieving instant feedback or alarm.

[0018] According to an embodiment of the present invention, the plurality of feature points have at least rotation invariance and translation invariance. In order to reduce the influence of displacement and rotation of vehicle components, according to the present invention, feature points with rotation invariance and translation invariance are preferably used to extract feature points of the template image and match feature points of the acquired image, thereby reducing the distortion of the corrected image.

[0019] According to one embodiment of the present invention, the multiple feature points are corner points, in particular Harris corner points or Shi-Tomasi corner points. Here, corner points can be understood as the connection points of the contour lines of an object. When the shooting angle changes, the corner points can still maintain stable properties well. Therefore, by using corner points as feature points, the amount of data of image information can be effectively reduced while retaining important features of the image graphics, the speed of calculation is improved, which is conducive to reliable matching of images and makes real-time processing possible. Here, Harris corner points or Shi-Tomasi corner points are preferably used, or SIFT algorithm, SURF algorithm or ORB algorithm are advantageously used for feature point matching.

[0020] According to one embodiment of the present invention, a convolutional neural network binary classifier is used to identify the presence of a part in the intercepted predetermined area, wherein the convolutional neural network binary classifier is trained by using an image with a part installed as a positive sample. Here, the binary classifier is used to determine whether the acquired image is a part that should be installed. Here, by using only binary classification processing to determine whether a part exists or not, the requirements for computing performance in the recognition process are reduced, and the recognition result is quickly obtained, so that the worker can be promptly warned at the installation station that the part to be installed is missing.

[0021] According to one embodiment of the present invention, a convolutional neural network multi-classifier is used to identify the type of parts in the intercepted predetermined area, wherein the convolutional neural network multi-classifier is trained by images with different types of parts installed. Since the reserved installation positions on vehicle components generally involve gaps of standard sizes, parts may be misinstalled during the production process, such as installing rivets where clips should be installed. These errors may also be difficult to detect during manual quality inspection. Here, a convolutional neural network multi-classifier is used to determine the type of installed parts, so that workers can be promptly warned at the installation station that the parts to be installed are misinstalled.

[0022] According to an implementation form of the present invention, the classifier uses a MobileNet neural network, thereby further reducing the requirements for computing performance during the recognition process, improving the computing speed, and thus quickly realizing the judgment of the existence of parts and / or the type of parts.

[0023] According to one embodiment of the present invention, the image of the rotatably supported vehicle component is subjected to grayscale transformation, in particular grayscale normalization. For general vehicle components and parts, there are no special provisions for color. The image can be advantageously subjected to grayscale transformation, thereby further reducing the amount of calculation and the requirements for computing performance compared to the processing of color images. In particular, the image can also be subjected to grayscale normalization to reduce the influence caused by differences in illumination.

[0024] According to one embodiment of the present invention, the vehicle component includes a rotatably supported vehicle instrument panel, a vehicle seat, a vehicle body, a vehicle chassis or a vehicle powertrain, etc.; and / or the parts include buckles, rivets, bolts, screws, nuts, pins, retaining clips or buttons, etc. The present invention is not limited to the listed vehicle components and parts, and can also be adapted to other locations and / or other parts in the vehicle according to specific needs.

[0025] A second aspect of the present invention relates to a computer-readable storage medium having executable instructions which, when executed, cause a computer to carry out the method according to the present invention for identifying a part mounted on a rotatably supported vehicle component.

[0026] A third aspect of the invention relates to a device for identifying a part mounted on a rotatably supported vehicle component, the device comprising:

[0027] - a memory storing computer executable instructions; and

[0028] - A processor configured to execute computer executable instructions, wherein the computer executable instructions, when executed by the processor, implement the method according to the invention for identifying a part mounted on a rotatably supported vehicle component.

[0029] A fourth aspect of the present invention relates to a system for identifying a part mounted on a rotatably supported vehicle component, the system comprising:

[0030] - an industrial camera for capturing an image of the rotatably supported vehicle component;

[0031] - a device according to the invention for detecting parts mounted on a rotatably mounted vehicle component;

[0032] - An alarm device for outputting an alarm signal when an error is detected.

[0033] The device and system according to the present invention can also automatically monitor the assembly process of rotatably supported vehicle parts in real time. The image can be quickly and accurately corrected and the parts to be installed can be identified with the lowest possible amount of calculation. As a result, the identified errors, such as missing or wrongly installed parts, can be promptly warned to the workers at the installation station, ensuring the correct installation of the parts on the vehicle parts, avoiding rework or repairs due to missing or wrong installation, and improving the quality of the vehicle. Without affecting the normal production of the original installation station, the manual operation of the production workers is reduced, the labor cost is reduced, and the production efficiency is improved.

[0034] According to one embodiment of the present invention, the system includes a data interface, which is configured to: obtain information about vehicle components and / or parts from a vehicle manufacturing system; and / or receive a signal that a vehicle component is in place from the vehicle manufacturing system; and / or send a corrected image and / or part recognition result to the vehicle manufacturing system. Here, information about vehicle components, such as vehicle component type, parts to be installed, etc., can be obtained from the vehicle manufacturing system to retrieve the corresponding template image and its feature points, and adaptively match the recognition process. A signal that a vehicle component is in place can be received from the vehicle manufacturing system to trigger the shooting of an industrial camera and the subsequent recognition process with a time delay. Corrected images and / or part recognition results can also be sent to the vehicle manufacturing system, such as whether the parts to be installed exist, whether the type of parts is correct, etc., so as to facilitate the archiving, querying, and tracing of quality data, and can also provide assistance for subsequent process optimization.

[0035] A fifth aspect of the invention relates to a vehicle production line comprising a system according to the invention for identifying parts mounted on a rotatably mounted vehicle component.

[0036] It should be noted that the features, functions, effects and advantages of one aspect of the present invention can also refer to the above description of other aspects of the present invention. In addition, the various aspects described in this document can be combined with each other in a variety of ways. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 A flow chart of the method according to the invention for identifying a component mounted on a rotatably mounted vehicle component is shown.

[0038] Figure 2 An acquired image of a rotatably mounted vehicle component is shown by way of example;

[0039] Figure 3a The template diagram used in the present invention is exemplarily shown;

[0040] Figure 3bThe exemplary diagram shows the position of the template from Figure 2 The image area directly captured in ;

[0041] Figure 4 The example shows that feature point matching is performed on the acquired image based on a plurality of feature points preset in the template image;

[0042] Figure 5 It is exemplarily shown that the acquired image is triangulated based on the matched feature points;

[0043] Figure 6 It is exemplarily shown that each of the divided triangular regions is mapped to a corresponding triangular region constructed by feature points preset in the template image by means of affine transformation, so as to obtain a corrected image;

[0044] Figure 7 exemplarily showing a predetermined area of ​​a part being cut out from the corrected image;

[0045] Figure 8 exemplarily showing identification of a part in the captured predetermined area; and

[0046] Fig. 9 A system according to the invention for detecting parts mounted on a rotatably mounted vehicle component is shown by way of example. DETAILED DESCRIPTION

[0047] Figure 1 A flow chart of a method 100 according to the invention for identifying a part mounted on a rotatably mounted vehicle component is shown. The method 100 comprises:

[0048] Step S1: acquiring an image of the rotatably supported vehicle component;

[0049] Step S2: performing feature point matching on the acquired image based on a plurality of feature points preset in the template image;

[0050] Step S3: triangulate the acquired image based on the matched feature points;

[0051] Step S4: mapping each of the divided triangular regions to corresponding triangular regions constructed by feature points preset in the template image by means of affine transformation, so as to obtain a corrected image, thereby overcoming image distortion and restoring the region of the part;

[0052] Step S5: intercepting a predetermined area of ​​the part from the corrected image;

[0053] Step S6: Identify parts in the captured predetermined area.

[0054] Figure 2 The captured image of a rotatably supported vehicle component is shown as an example. Here, a vehicle instrument panel is used as an example to illustrate a rotatably supported vehicle component, and a buckle is used as an example of a part to be installed. The vehicle instrument panel is supported on a bracket on a production line guide rail so as to be rotatable about an axis. Figure 2 As shown, the vehicle dashboard is flipped to a position with the bottom facing upward. The positions of the buckles to be installed are marked with four circles in the figure. As mentioned above, the plastic seal fixed by the buckle will play a vital role in the drainage of the air conditioning duct. Preferably, the image of the rotatably supported vehicle component is gray-scale transformed, especially gray-scale normalized.

[0055] Figure 3a The template image used in the present invention is shown as an example. Here, the template image advantageously only relates to the image region containing the parts, thereby simplifying the subsequent image processing and calculation. Figure 3b The exemplary diagram shows the position of the template from Figure 2 The image area is captured directly from . Figure 3a and 3b The corresponding positions are marked with boxes. Figure 3a and 3b It is clear that the image area containing parts is deformed due to the different support angles of the instrument panel. The reasons can be attributed to the worker's adjustment, the offset of the support point, the size error of the bracket, the force applied during the installation process, etc. Figure 3b As shown in the figure, the two lower clips are not completely in the position where the clips should appear in the template image. Figure 3b Such images are also difficult to correct by simple translation, rotation or scaling, which makes it impossible to accurately locate them in the image and difficult to perform conventional template matching.

[0056] Figure 4 The example shows that the feature points of the acquired image are matched based on a plurality of feature points preset in the template image. Figure 4 The upper left corner shows Figure 3a The template diagram shown in Figure 4 The right side shows Figure 2The image of the vehicle dashboard obtained is shown. The corresponding feature points are marked in these two images respectively, and the two corresponding feature points are connected by a line segment to indicate the corresponding relationship. In the present invention, it can be stipulated that the multiple feature points have at least rotation invariance and translation invariance. Feature point extraction of the template image is performed in advance by using feature points that have at least rotation invariance and translation invariance, and feature point matching of the acquired image is performed with such feature points. Here, it is preferred to use corner points as the multiple feature points. Here, the corner points can be understood as the connecting points of the contour lines of the object, and when the shooting angle changes, the corner points can still maintain stable properties well. The corner points are in particular Harris corner points or Shi-Tomasi corner points. In addition, the SIFT algorithm, SURF algorithm or ORB algorithm can also be advantageously used for feature point matching.

[0057] Figure 5 The triangulation of the acquired image based on the matched feature points is shown as an example, and in particular, Delaunay triangulation is implemented, thereby segmenting the image area involving the feature points in the acquired image into a number of triangular areas.

[0058] Figure 6 The example shows that each of the divided triangular regions is mapped to the corresponding triangular regions constructed by the preset feature points in the template image by means of affine transformation to obtain a corrected image, thereby overcoming the image distortion and restoring the region of the part. Figure 6 The left figure shows Figure 5 The image area triangulated in the middle figure shows the corresponding triangular areas constructed by the preset feature points in the template image, and the right figure shows the image processing result of mapping each triangular area in the left figure to each triangular area in the middle figure by means of affine transformation, that is, the corrected image is obtained. Figure 3b The area shown is restored to the effect of the module diagram without distortion. Here, localized deformation matching is performed for the image status of the rotatably supported vehicle parts to overcome the image distortion caused by the worker's adjustment, the offset of the support point, the size error of the bracket, the force applied during the installation process, etc., and more accurately correct it to the state of the template diagram.

[0059] Figure 7 The predetermined area of ​​the parts cut out from the corrected image is shown as an example. Here, the predetermined area is the area where each part is marked in the template image. Figure 7 It can be seen that the positions of each part in the corrected image can correspond well to the template image, so each buckle can be accurately positioned based on it.

[0060] Figure 8 The identification of parts in the intercepted predetermined area is shown as an example. Figure 7 The four intercepted predetermined regions of the parts shown in are identified respectively. Exemplarily, a convolutional neural network ten classifier is used to identify ten different parts or nine different parts and one part not installed in the intercepted predetermined region, wherein the convolutional neural network ten classifier is trained by the ten images.

[0061] like Figure 8 An exemplary convolutional neural network is shown. The convolutional neural network contains eight layers in total, which are distributed according to the input layer, convolution layer C1, downsampling layer or pooling layer S2, convolution layer C3, downsampling layer or pooling layer S4, fully connected layer C5, fully connected layer F6, and output layer. A 5*5 convolution kernel is used here. Based on the affine invariance brought about by the local perception and weight sharing of the convolutional neural network, it can further eliminate the interference caused by the rotation of vehicle parts and part installation errors, and further accurately identify the existence of parts and / or the type of parts. Not limited to Figure 8 In the example shown, the classifier may also preferably adopt a MobileNet neural network.

[0062] Fig. 9 A system 200 for identifying parts mounted on a rotatably supported vehicle component according to the present invention is shown by way of example. The system 200 comprises:

[0063] - an industrial camera 201 for capturing an image of the rotatably supported vehicle component;

[0064] - a device 202 according to the invention for identifying a part mounted on a rotatably mounted vehicle component;

[0065] - An alarm device 203, for outputting an alarm signal when an error is identified.

[0066] Here, the device 202 for identifying parts installed on a rotatably supported vehicle component includes: a memory storing computer-executable instructions; and a processor configured to execute the computer-executable instructions, wherein when the computer-executable instructions are executed by the processor, the method of identifying parts installed on a rotatably supported vehicle component of the present invention is implemented.

[0067] Fig. 9 2 shows an exemplary case where a vehicle instrument panel is rotatably supported on a bracket. Here, the vehicle instrument panel is shown in a state where it has not been rotated, but it can be rotated according to the installation requirements. Fig. 9The axis indicated by the dotted line is flipped to facilitate the workers to install the parts. In the present invention, the industrial camera 201 will take an image of the rotatably supported vehicle component after the worker has completed the installation, especially when the worker leaves the workstation to pick up the component again. In this way, the assembly process of the rotatably supported vehicle component can be monitored in real time.

[0068] Advantageously, the system 200 may also include a data interface, which is configured to: obtain information about vehicle components and / or parts from the vehicle manufacturing system IPS; and / or receive a signal that the vehicle components are in place from the vehicle manufacturing system IPS; and / or send a corrected image and / or part recognition result to the vehicle manufacturing system IPS. Here, the system 200 can obtain information about vehicle components, such as vehicle component types, parts to be installed, etc., through communication with the vehicle manufacturing system IPS, so as to retrieve the corresponding template image and its feature points, and perform adaptive matching on the recognition process, etc.; trigger the shooting of the industrial camera and the subsequent recognition process with a delay through the signal that the vehicle components are in place; and facilitate the archiving, querying and tracing of quality data.

[0069] The present invention may also be a computer program product. The computer program product may include a computer-readable storage medium having computer-readable program instructions thereon for causing a processor to execute various aspects of the present invention.

[0070] A computer readable storage medium may be a tangible device that can retain and store instructions for use by an instruction execution device. A computer readable storage medium may be, for example but not limited to, an electronic storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing.

[0071] The computer readable program instructions may execute entirely on the terminal's computer, partly on the terminal's computer, as a stand-alone software package, partly on the terminal's computer and partly on a remote computer, or entirely on a remote computer or server.

[0072] This paper describes various aspects of the present invention with reference to the flow chart and / or block diagram of the method, system and computer program product according to embodiments of the present invention.It should be understood that each box in the flow chart and / or block diagram and the combination of boxes in the flow chart and / or block diagram can be implemented by computer readable program instructions.

[0073] The present invention is not limited to the embodiments shown, but includes or extends to all technical equivalents that may fall within the effective scope of the appended claims. The position descriptions selected in the specification, such as top, bottom, left, right, etc., refer to the direct description and the drawings shown and can be transferred to the new position according to the meaning when the position changes.

[0074] The features disclosed in the present application may be essential and may be implemented both individually and in any combination for the realization of the exemplary embodiments in different configurations.

[0075] Although some aspects are described in association with devices, it should be understood that these aspects are also descriptions of corresponding methods, so that a device of a module or a device of a system can also be understood as a corresponding method step or a feature of a method step. Similarly, aspects described in association with a method step or as a method step are also descriptions of corresponding modules or details or features of corresponding devices.

[0076] Therefore, a computer-readable storage medium may be machine-readable or computer-readable. Therefore, in some embodiments, a computer-readable storage medium includes a data carrier having executable instructions that can cooperate with a programmable computer system or a programmable hardware component to implement one of the methods described herein. Therefore, one embodiment is a data carrier, a digital storage medium, or a computer-readable storage medium, on which a program for implementing one of the methods described herein is recorded.

[0077] Furthermore, another embodiment is a data stream, a signal sequence or a signal sequence, which is a program for implementing one of the methods described herein. The data stream, signal sequence or signal sequence can be configured, for example, for transmission via a data communication connection, for example via the Internet or another network. Thus, an embodiment can also be a signal sequence representing data, which is suitable for transmission via a network or a data communication connection, wherein the data is a program.

[0078] Although the present invention has been disclosed as above in the form of preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art may make possible changes and modifications to the technical solutions of the present invention by utilizing the methods and technical contents disclosed above without departing from the spirit and scope of the present invention. Therefore, any simple modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the contents of the technical solutions of the present invention shall fall within the protection scope of the technical solutions of the present invention.

Claims

1. A method for identifying a part mounted on a rotatably supported vehicle component, the method comprising: - acquiring an image of the rotatably supported vehicle component; - Matching feature points of the acquired image based on multiple feature points preset in the template image; -Triangulate the acquired image based on the matched feature points; - Mapping each of the divided triangular regions to the corresponding triangular regions constructed by the preset feature points in the template image by means of affine transformation to obtain a corrected image, thereby overcoming the image distortion and restoring the region of the part; - extracting a predetermined area of ​​the part from said corrected image; - Identify the part in the intercepted predetermined area.

2. The method according to claim 1, characterized in that The template image only relates to the image area containing the parts.

3. The method according to claim 1, characterized in that The plurality of feature points have at least rotation invariance and translation invariance.

4. The method according to claim 3, characterized in that The multiple feature points are corner points; or the feature points are matched using a SIFT algorithm, a SURF algorithm or an ORB algorithm.

5. The method according to claim 4, characterized in that The corner points are Harris corner points or Shi-Tomasi corner points.

6. The method according to any one of claims 1 to 5, characterized in that A convolutional neural network binary classifier is used to identify the presence of parts in the intercepted predetermined area, wherein the convolutional neural network binary classifier is trained using images with parts installed as positive samples.

7. The method according to any one of claims 1 to 5, characterized in that A convolutional neural network multi-classifier is used to identify the type of parts in the intercepted predetermined area, wherein the convolutional neural network multi-classifier is trained by images with different types of parts installed.

8. The method according to claim 6, characterized in that The convolutional neural network binary classifier adopts the MobileNet neural network.

9. The method according to claim 7, characterized in that The convolutional neural network multi-classifier adopts the MobileNet neural network.

10. The method according to any one of claims 1 to 5, characterized in that A grayscale conversion is performed on the image acquired of the rotatably mounted vehicle component.

11. The method according to claim 10, characterized in that The grayscale transformation is grayscale normalization.

12. The method according to any one of claims 1 to 5, characterized in that The vehicle component comprises a rotatably supported vehicle instrument panel, a vehicle seat, a vehicle body part, a vehicle chassis or a vehicle powertrain; and / or The parts include snaps, rivets, bolts, screws, nuts, pins, retaining clips or buttons.

13. A computer-readable storage medium having executable instructions, which, when executed, cause a computer to perform the method according to any one of claims 1 to 12.

14. A device for identifying a part mounted on a rotatably supported vehicle component, the device comprising: - memory storing computer executable instructions; as well as - A processor configured to execute computer executable instructions, wherein the computer executable instructions, when executed by the processor, implement the method according to any one of claims 1 to 12.

15. A system for identifying a part mounted on a rotatably supported vehicle component, the system comprising: - an industrial camera for capturing an image of the rotatably supported vehicle component; - a device for detecting a part mounted on a rotatably mounted vehicle component according to claim 14; - An alarm device for outputting an alarm signal when an error is detected.

16. The system according to claim 15, characterized in that The system comprises a data interface configured to: Obtaining information about vehicle components and / or parts from vehicle manufacturing systems; and / or receiving a signal from a vehicle manufacturing system that a vehicle component is in place; and / or The corrected image and / or part recognition result is sent to the vehicle manufacturing system.

17. The system according to claim 15 or 16, characterized in that The vehicle component comprises a rotatably supported vehicle instrument panel, a vehicle seat, a vehicle body part, a vehicle chassis or a vehicle powertrain; and / or The parts include snaps, rivets, bolts, screws, nuts, pins, retaining clips or buttons.

18. Vehicle production line comprising a system for identifying parts mounted on a rotatably mounted vehicle component according to one of claims 15 to 17.

Images