Method, storage medium, device and system for identifying strip trim of a vehicle

Abstract

The present invention relates to a method for identifying a strip-shaped decorative part of a vehicle, the method comprising: acquiring an image of a specified area of ​​the strip-shaped decorative part; preprocessing the acquired image to obtain a binary image; performing edge straight line segment detection on the preprocessed binary image; extracting edge straight line segments that roughly conform to the specified direction of the strip-shaped decorative part from the detected edge straight line segments; fusing the extracted edge straight line segments to obtain fused line segments; performing curve fitting on the fused line segments; screening out a curve related to the strip-shaped decorative part according to the result of the curve fitting; and judging the existence of the strip-shaped decorative part based on the geometric features of the strip-shaped decorative part. The present invention also relates to a computer-readable storage medium, a device for identifying a strip-shaped decorative part of a vehicle, a system for identifying a strip-shaped decorative part of a vehicle, and a vehicle production line including the system. The present invention can automatically and simply and reliably identify strip-shaped decorative parts, thereby replacing manual inspection.

Description

Technical Field

[0001] The present invention relates to a method for identifying a strip-shaped trim of a vehicle, a computer-readable storage medium, an apparatus for identifying a strip-shaped trim of a vehicle, a system for identifying a strip-shaped trim of a vehicle, and a vehicle production line comprising the system. Background Art

[0002] With the evolution of the automobile industry, the user's needs have gradually upgraded, and the requirements for vehicle personalization and comfort have gradually increased, especially paying more attention to the decorative texture of the vehicle and the environmental experience when driving. In recent years, light strips or ambient lights used to decorate cars and create atmosphere, as well as metal strips, colored strips and wooden strips used to enhance the sense of design and texture have been used in more and more car models. In particular, based on the personalized needs of users, different strips of the above-mentioned decorative parts may need to be assembled at different positions of the vehicle on a vehicle production line.

[0003] At present, the inspection of such strip-shaped decorative parts can only be completed manually by quality inspectors based on paper work orders or electronically displayed assembly information, so it is still difficult to avoid missing parts. If the missing parts are not handled in time and are not discovered until the sales or after-sales stage, it will greatly reduce customer satisfaction with the overall brand. In addition to wasting extra time and energy on repairs, it will also have a great negative impact on the overall brand reputation. Summary of the invention

[0004] The object of the present invention is to provide a possibility for simple and reliable automatic identification of trim strips in vehicle manufacturing, thereby replacing traditional manual inspection, quickly and accurately identifying trim strips with the lowest possible computational effort, reducing labor costs and improving production efficiency.

[0005] In order to solve the above-mentioned task, the present invention provides a method, a computer-readable storage medium, a device, a system for identifying a strip trim of a vehicle, and a vehicle production line including the system.

[0006] A first aspect of the present invention relates to a method for identifying a strip trim of a vehicle, the method comprising:

[0007] - acquiring an image of a specified area of ​​the strip-shaped trim;

[0008] - preprocessing the acquired image to obtain a binary image;

[0009] - Implement edge straight line segment detection on the pre-processed binary image;

[0010] - extracting edge straight line segments that roughly conform to the prescribed direction of the strip-shaped decorative part from the detected edge straight line segments;

[0011] - Fusing the extracted edge straight line segments to obtain fused line segments;

[0012] - performing curve fitting on the fused line segments and selecting curves related to the strip-shaped decorative parts according to the fitting results;

[0013] - Determining the presence of a strip of trim based on its geometric features.

[0014] In vehicles, strip-shaped decorative parts such as light strips, metal strips, colored strips and wooden strips are usually bonded, anchored, locked or embedded in vehicle parts such as dashboards, doors and bodies, so that they fit on the surface or edges of vehicle parts. This makes the assembled strip-shaped decorative parts mostly fit on the surface or edges of streamlined industrially designed vehicle parts, thereby enhancing the visual perception. Therefore, these strip-shaped decorative parts usually have a certain curvature and extend along a curve when assembled, rather than stretching straight. Moreover, in order to highlight the visual perception, the strip-shaped decorative parts are also different from the vehicle parts to which they are installed in terms of color and brightness.

[0015] Based on the above background, the present invention provides a simple, reliable and low-computation method for identifying strip-shaped decorative parts.

[0016] In order to identify the strip-shaped decorative parts, an image of the area where the strip-shaped decorative parts should be provided may be first acquired. The acquired image may be a grayscale image or a color image, in particular an RGB image.

[0017] The acquired image is then preprocessed to obtain a binary image. Here, the preprocessing may include not only binary conversion to generate the desired binary image, but also image capture, scaling, rotation, adjustment of brightness, contrast, resolution and saturation, etc. Here, in particular, an appropriate threshold may also be specified to obtain a binary image that can well reflect the characteristics of the strip-shaped decorative piece. Particularly preferably, in order to simplify subsequent calculations, the image may be rotated according to the direction of the strip-shaped decorative piece to be identified, so that the strip-shaped decorative piece extends substantially horizontally or vertically in the rotated image.

[0018] Next, edge straight line segment detection is performed on the pre-processed binary image. Here, edge straight line segment detection in the binary image can be performed based on gradient, so as to extract each line segment extending in a straight line direction at the black and white boundary in the binary image. The edge straight line segment detected here can at least represent the edge of the strip trim, but may also include the edge, gap, texture or contour of other vehicle parts in the image.

[0019] Then, the edge straight line segments that roughly match the specified direction of the strip-shaped decorative part are extracted from the detected edge straight line segments. Here, based on the fact that the strip-shaped decorative part still maintains its approximate extension direction and only has a certain curvature when attached to the surface or edge, the edge straight line segments that roughly match the specified direction of the strip-shaped decorative part can be extracted by specifying a certain angle relative to its extension direction. Taking the light strip attached to the bottom of the vehicle dashboard as an ambient light as an example, the edge straight line segments with an inclination angle within ±30° relative to the horizontal direction can be selected, so as to filter out other interfering edge straight line segments in the image while considering the curvature.

[0020] Then, the extracted edge straight line segments can be fused to obtain fused line segments. Due to interfering factors such as image noise or bending of strip trims, the selected edge straight line segments are not connected to each other but may be discontinuous. In order to associate the edge straight line segments representing the same edge with each other, these edge straight line segments can be fused according to certain fusion conditions. For example, the fusion conditions of the edge straight line segments can be the thresholds of the absolute distance and relative distance between the endpoints of the two edge straight line segments, the parallelism of the extension directions of the two edge straight line segments, etc. In addition, the fused line segments whose lengths after fusion match or meet the standards can be screened out based on the total length of the line segments to further exclude the edges, textures, gaps or contours of other vehicle components.

[0021] Next, curve fitting is performed on the fused line segments; and the curves related to the strip-shaped trim are screened out based on the curve fitting results. Since the acquired image may also contain the edges, textures, gaps or contours of other vehicle components, it is necessary to screen out the curves related to the strip-shaped trim. To this end, curve fitting is performed on the fused line segments to obtain the mathematical expression of the fitting curve of each fused line segment, and the edges, textures, gaps or contours of other vehicle components detected in the image are excluded based on the fitting results. In this way, interference lines in the image are excluded and deviations caused by the positioning of vehicle components and image angles are reduced.

[0022] Finally, the presence or absence of the strip-shaped decorative part is determined based on the geometric features of the strip-shaped decorative part. The presence or absence of the strip-shaped decorative part is determined by comparing the geometric features of the fused line segment or the fitting curve with the geometric features of the strip-shaped decorative part in the assembled state. To this end, a certain threshold range is predefined for the geometric features, and the geometric features calculated based on the acquired image are compared with the threshold range to obtain the recognition result of the strip-shaped decorative part.

[0023] According to the present invention, a computer-implemented method for identifying strip trims of a vehicle is provided, which can identify the strip trims to be installed through an image processing process designed for the installation of strip trims in a vehicle, quickly and accurately identify the strip trims with as low a calculation amount as possible, and does not impose high requirements on the computing performance of the computer. Therefore, the present invention can be deployed in a vehicle production line at a low cost, effectively saving the labor cost originally required, avoiding omissions in manual inspection, and optimizing the production process.

[0024] According to an embodiment of the present invention, the geometric features of the strip-shaped decorative piece may include, for example, the width, length and / or curvature of the strip-shaped decorative piece. Preferably, the distance between the fitting curves (which represent the opposite edges of the strip-shaped decorative piece) with similar distances, in particular the average distance, the distance range or the clamped pixel value distribution, etc., may be calculated; or the distance between the fusion curves with similar distances, in particular the average distance, the distance range or the clamped pixel value distribution, etc. may be calculated. Such calculated geometric features are compared with a predetermined threshold range to determine whether the strip-shaped decorative piece exists or not.

[0025] According to an embodiment of the present invention, a quadratic function fitting based on the least squares method is performed in the curve fitting. Here, since the fitting of the strip-shaped decorative parts usually only presents a slight curvature, it is advantageous to use a quadratic function to fit the edge trend of the strip-shaped decorative parts, thereby simplifying the calculation, improving the calculation speed and reducing the computing power requirement. Here, the fitted quadratic function can be expressed as follows:

[0026] y=ax 2 +bx+c.

[0027] According to an embodiment of the present invention, the fitting curve is screened according to the quadratic term coefficient of the fitted quadratic function. According to the above formula, the quadratic term coefficient of the quadratic function is a, and its role in the quadratic function is to determine the bending direction and degree of the function graph. Therefore, the edge of the strip-shaped decorative part to be identified can be simply and accurately distinguished from other edges in the acquired image by the quadratic term coefficient.

[0028] According to an embodiment of the present invention, the fused line segment is sampled at intervals, and each sampling point is input into the curve fitting. Here, in order to simplify the fitting process, reduce the amount of calculation and improve the calculation rate, the fused line segment is sampled at intervals, and only some points in the fused line segment are sampled to implement the curve fitting.

[0029] According to an embodiment of the present invention, equally spaced sampling is performed on the fused line segment based on a predetermined sampling interval; or equally spaced sampling is performed on each edge straight line segment included in the fused line segment based on a predetermined sampling interval; or interval sampling is performed based on the length of each edge straight line segment included in the fused line segment. This embodiment specifies different situations for interval sampling: equally spaced sampling is performed on the entire fused line segment, wherein the sampling points are evenly distributed on the entire fused line segment; equally spaced sampling is performed on each line segment before fusion, thereby taking into account the position information contained in each line segment before fusion; interval sampling is performed on each line segment before fusion according to its different lengths, thereby better retaining the information at the bend or especially at the inflection point.

[0030] According to one embodiment of the present invention, in the implementation of interval sampling based on the length of each edge straight line segment included in the fused line segment, the number and / or interval of sampling points for each edge straight line segment is determined according to the piecewise function of the length. In this embodiment, the piecewise function of the length is predetermined, so that the number of sampling points and / or the interval of adjacent sampling points are simply specified according to the different lengths of the edge straight line segments before fusion. Thereby, it is still guaranteed that very short line segments can be sampled and input into the curve fitting. For long line segments with relatively stable trends, fewer sampling points and / or larger intervals are specified. This can simplify the fitting, reduce the amount of calculation and increase the calculation rate without losing important information for the fitting.

[0031] According to an embodiment of the present invention, the edge straight line segment detection utilizes LSD line segment detection or Hough transform line segment detection. Here, in order to further improve the calculation rate, LSD line segment detection is preferably used to detect edge straight line segments in the pre-processed binary image.

[0032] According to one embodiment of the present invention, before the edge straight line segment detection is implemented, a binary classifier trained with an image of a specified area containing at least a complete strip of decorative parts as a positive sample is used to determine whether the acquired image contains a valid image of the specified area of ​​the strip of decorative parts. In an actual vehicle production line, an operator may pass by the inspection station for photographing the strip of decorative parts, thereby causing interference. In order to achieve the coordinated operation of the recognition method according to the present invention and the worker's operation, before the edge straight line segment detection is implemented, the binary classifier is used to determine whether the acquired image contains a complete image of the specified area of ​​the strip of decorative parts. Here, in the present invention, the judgment of image integrity only uses binary classification processing, and no object recognition is performed in the image, thereby reducing the requirements for computing performance in the recognition process. Particularly preferably, the binary classifier uses a MobileNet convolutional neural network. As a result, the requirements for computing performance in the recognition process are further reduced, the computing speed is improved, and the judgment of whether the complete image of the specified area of ​​the strip of decorative parts is quickly realized.

[0033] According to one embodiment of the present invention, the strip-shaped decorative part is a light strip, a metal side strip, a color side strip or a wooden side strip. Here, the light strip is usually installed on the steering wheel, center console, instrument panel, roof, door, trunk, headlight and other positions of the vehicle as an ambient light. Metal side strips or color side strips are usually installed on the steering wheel, center console, instrument panel, roof, door, trunk, headlight and other positions of the vehicle. And wooden side strips are usually installed on the steering wheel, center console, instrument panel and other positions of the vehicle as interior decoration parts.

[0034] A second aspect of the present invention relates to a computer-readable storage medium having executable instructions which, when executed, cause a computer to perform a method according to the present invention.

[0035] A third aspect of the present invention relates to a device for identifying a strip-shaped trim of a vehicle, the device comprising:

[0036] - a memory storing computer executable instructions; and

[0037] - a processor configured to execute computer executable instructions, wherein said computer executable instructions, when executed by said processor, implement the method according to the present invention.

[0038] A third aspect of the present invention relates to a system for identifying a strip trim of a vehicle, the system comprising:

[0039] - an industrial camera for capturing an image of a defined area of ​​the strip;

[0040] - a device according to the invention for detecting a strip of trim on a vehicle;

[0041] - an alarm device for outputting an alarm signal when it is determined that the belt-shaped fastener is not present.

[0042] According to one embodiment of the present invention, the industrial camera is set at an unmanned position. The unmanned position can be understood as any position after the assembly station for installing strip-shaped decorative parts on the vehicle production line where few people pass by, especially the waiting area before entering the next station. Here, the industrial camera can be set up above, below or to the side of the unmanned position to capture images of the specified area with strip-shaped decorative parts. In this way, the strip-shaped decorative parts can be identified without changing the original assembly station and without adding additional work processes for production workers.

[0043] According to one embodiment of the present invention, the system comprises a data interface, and the data interface is configured to: obtain information about the strip trim from the vehicle manufacturing system; and / or send a strip trim identification result to the vehicle manufacturing system.

[0044] 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

[0045] Figure 1 A flow chart showing a method for identifying a strip-shaped trim of a vehicle according to the present invention;

[0046] Figures 2a to 2h Schematic diagrams showing the processing of images according to the individual method steps of the method for identifying a strip-shaped trim of a vehicle according to the present invention;

[0047] Figure 3a A schematic diagram showing interval sampling in curve fitting;

[0048] Figure 3b Schematic diagram showing the curve fitting results.

[0049] Figure 4 A schematic diagram of a system for recognizing a trim strip of a vehicle according to the present invention is shown. DETAILED DESCRIPTION

[0050] Figure 1 A flow chart of a method for identifying a strip trim of a vehicle according to the present invention is shown. The method 100 for identifying a strip trim of a vehicle comprises:

[0051] Step S1: acquiring an image of a specified area of ​​a strip-shaped decorative part;

[0052] Step S2: preprocessing the acquired image to obtain a binary image;

[0053] Step S3: performing edge straight line segment detection on the pre-processed binary image;

[0054] Step S4: extracting edge straight line segments that roughly conform to the prescribed direction of the strip-shaped decorative part from the detected edge straight line segments;

[0055] Step S5: fusing the extracted edge straight line segments to obtain fused line segments;

[0056] Step S6: performing curve fitting on the fused line segment;

[0057] Step S7: selecting a curve related to the strip-shaped decorative part according to the curve fitting result;

[0058] Step S8: determining the existence of the strip-shaped decorative part based on the geometric features of the strip-shaped decorative part.

[0059] The following comparison Figures 2a to 2h The processing of the image in each method step of the method for identifying a strip-shaped trim of a vehicle according to the present invention is described in more detail. Here, a light strip of an ambient light installed on a dashboard is used as an example of a strip-shaped trim. Figure 2a A white strip is shown in the middle of the right side. Not limited to the example shown, the strip-shaped decorative piece can also be a metal edge strip, a colored edge strip or a wooden edge strip.

[0060] After the method 100 of the present invention is started, an image of a specified area of ​​the strip trim, that is, an image of the area where the strip trim is to be provided, is first acquired according to step S1. The image is shown in FIG2 , and can be a grayscale image or a color image, in particular an RGB image.

[0061] Next, the acquired image is preprocessed according to step S2 to obtain a binary image. Figure 2b The black and white image shown. Here, the preprocessing may also include image capture, scaling, rotation, adjustment of brightness, contrast, resolution and saturation, etc. In particular, an appropriate threshold may be specified for binarization so that the obtained binarized image well reflects the characteristics of the strip-shaped decorative part.

[0062] Then, according to step S3, edge straight line segment detection is performed on the pre-processed binary image. The edge straight line segment detection can be performed using LSD line segment detection or Hough transform line segment detection. Figure 2c The results of edge straight line segment detection are schematically shown in FIG. Figure 2b Different grayscale lines are used to schematically represent the identified edge straight line segments.

[0063] Then, according to step S4, the edge straight line segments that roughly conform to the prescribed direction of the strip-shaped decorative part are extracted from the detected edge straight line segments. Figure 2d As shown, since the light bar extends substantially horizontally, only edge straight line segments with an inclination angle within ±30° relative to the horizontal direction are selected, so as to filter out other interfering edge straight line segments in the image while taking the curvature into consideration.

[0064] Next, according to step S5, the extracted edge straight line segments are fused to obtain fused line segments. Figure 2e As shown, the fused lines are represented by bold grayscale lines. Here, in order to show it more clearly, the fused line segments are shown on the acquired original image, wherein three grayscale lines are used to mark the fused line segments: the fused line segment indicating the upper edge of the light bar, the fused line segment indicating the lower edge of the light bar, and the fused line segment indicating the top edge of the instrument panel. Thereby, the edge straight line segments representing the same edge are associated with each other, and these edge straight line segments can be fused here according to certain fusion conditions. For example, the fusion condition of the edge straight line segments can be the threshold of the absolute distance and relative distance between the endpoints of the two edge straight line segments, the parallelism of the extension direction of the two edge straight line segments, etc. It is also preferred to screen the fused line segments whose lengths match or meet the standards again based on the length.

[0065] Then, according to step S6, curve fitting is performed on the fused line segments to obtain the fitted mathematical expressions of each fused line segment. Preferably, quadratic function fitting based on the least squares method is performed in the curve fitting. Further preferably, the fitting curve is screened according to the quadratic term coefficient of the fitted quadratic function. Figure 2f As shown, for a clearer illustration, the fused line segments and the fitting curve are shown on the original image, wherein, for a clear identification, a thinner grayscale line identifying the fitting curve passes through the original fused line and extends over the width of the entire image.

[0066] Then, according to step S7, the curves related to the strip-shaped decorative parts are selected according to the curve fitting results. Figure 2g As shown, the fitting curve indicating the upper edge of the light bar, the fitting curve indicating the lower edge of the light bar, and the fitting curve indicating the top edge of the instrument panel are shown in the Cartesian coordinate system. As mentioned above, based on the quadratic function y=ax 2 +bx+c to fit each fused line segment. Here, the fitting curve indicating the top edge of the instrument panel can be excluded according to the quadratic term coefficient a of the curve fitting. Specifically, the value range of the quadratic term coefficient a can be pre-set to [min_a, max_a]. It can be known from the calculation that the quadratic term coefficient of the fitting curve indicating the top edge of the instrument panel is not within the value range, so the fitting curve indicating the top edge of the instrument panel is excluded.

[0067] Finally, according to step S8, the presence of the strip-shaped decorative part is determined based on the geometric features of the strip-shaped decorative part. Figure 2hIt can be seen that the fitting curve indicating the upper edge of the light bar and the fitting curve indicating the lower edge of the light bar are approximately the same distance in the position of the light bar, that is, on the right half of the horizontal axis. This is because the width of the light bar is consistent in its longitudinal extension direction. Therefore, this geometric feature can be used to determine the existence of the light bar. Not limited to this, the length of the strip-shaped decorative part or its curvature in the installed state can also be considered.

[0068] In method 100, it is also preferred that before performing edge straight line segment detection, a convolutional neural network binary classifier trained with a specified area image containing at least a complete strip of decorative parts as a positive sample is used to determine whether the acquired image contains a valid image of the specified area of ​​the strip of decorative parts. Particularly preferably, the binary classifier uses a MobileNet neural network. Here, the classification of the image can be based on the acquired original image, but a pre-processed grayscale image or a binary image can also be used, thereby further simplifying the classifier and reducing the amount of calculation.

[0069] Figure 3a A schematic diagram showing the interval sampling in curve fitting, Figure 3b A schematic diagram showing the curve fitting result. In step S6: in curve fitting of the fused line segment, in order to simplify fitting, reduce the amount of calculation and increase the calculation rate, interval sampling can be implemented on the fused line segment, and each sampling point can be input into the curve fitting. According to the present invention, equal-interval interval sampling can be implemented on the entire fused line segment based on a predetermined sampling interval; or equal-interval interval sampling can be implemented on each edge straight line segment included in the fused line segment based on a predetermined sampling interval, or interval sampling can be implemented based on the length of each edge straight line segment included in the fused line segment. In particular, in the implementation of interval sampling based on the length of each edge straight line segment included in the fused line segment, the number and / or interval of sampling points for each edge straight line segment can also be determined based on a piecewise function of the length.

[0070] like Figure 3a As shown, the line segments L1, L2, and L3 before fusion are sampled at equal intervals. Here, 5, 6, and 10 sampling points are sampled for the line segments L1, L2, and L3 respectively. This allows the position information contained in each line segment before fusion to be taken into account without introducing errors due to the implemented line segment fusion. Here, the horizontal coordinate is advantageously used to calculate the intervals between the sampling points, thereby eliminating additional geometric calculations and the sampled values ​​are more convenient for subsequent curve fitting. Here, as shown in FIG. Figure 3a As shown, since the endpoints of a line segment often represent a change in the extension direction, it is further preferred to sample the endpoints of each line segment.

[0071] exist Figure 3b Shown in Figure 3aThe curve fitting results of each sampling point in . Here, the fitted curve is shown as curve L4. It can be seen from the figure that the distribution of each sampling point basically conforms to the trend of the quadratic function.

[0072] Figure 4 A schematic diagram of a system 200 for identifying a strip trim of a vehicle according to the present invention is shown. The system 200 comprises:

[0073] - an industrial camera 201 for capturing an image of a specified area with a strip-shaped trim;

[0074] - a device 202 for identifying a strip of trim on a vehicle;

[0075] - An alarm device 203 is used to output an alarm signal when it is determined that the belt-shaped fastener does not exist.

[0076] Here, the device 202 for identifying strip trims of a vehicle 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 for identifying strip trims of a vehicle according to the present invention is implemented.

[0077] like Figure 4 As shown, the system is exemplarily used to inspect instrument panels assembled on a vehicle production line to identify whether strip decorations to be installed thereon exist. The strip decorations can be light strips of ambient lights on the instrument panel or metal side strips, colored side strips or wooden side strips installed on the surface of the instrument panel. The industrial camera 201 is set at an unmanned position, especially above the instrument panel to be inspected, to capture images of a specified area with strip decorations. The system also includes a data interface, which is configured to obtain information about strip decorations from a vehicle manufacturing system 204 and / or send strip decoration recognition results to the vehicle manufacturing system 204.

[0078] The present invention is not limited to the embodiments shown, but includes or extends to all technical equivalents that fall within the effective scope of the attached claims. The features disclosed in the present application document can be implemented not only individually but also in any combination. In addition, it should be noted that the various drawings of the present invention are schematic and may not be shown to scale. In various embodiments, the methods, devices, equipment, systems and vehicle production lines are also not limited to the examples shown. The situations listed in the specification are only exemplary reference situations. In the specific application of the present invention, other implementation methods, process sequences, layout structures and connection relationships can be designed as needed.

Claims

1. A method for identifying a strip trim of a vehicle, the method comprising: - acquiring an image of a specified area of ​​the strip-shaped trim; - preprocessing the acquired image to obtain a binary image; - performing edge straight line segment detection on the pre-processed binary image, wherein the edge straight line segment detection uses LSD line segment detection or Hough transform line segment detection; - extracting edge straight line segments that conform to the specified direction of the strip trim from the detected edge straight line segments; - Fusing the extracted edge straight line segments to obtain fused line segments; - performing curve fitting on the fused line segment, and performing quadratic function fitting based on the least square method in the curve fitting; - Filter out the curve related to the strip decoration according to the curve fitting result; - Determining the presence of a strip of trim based on its geometric features.

2. The method according to claim 1, characterized in that The geometrical characteristics of the strip-shaped decoration include the width, length and / or curvature of the strip-shaped decoration.

3. The method according to claim 1, characterized in that The fitting curve is selected according to the quadratic term coefficient of the fitted quadratic function.

4. The method according to any one of claims 1 to 3, characterized in that Interval sampling is performed on the fused line segment, and each sampling point is input into the curve fitting.

5. The method according to claim 4, characterized in that - performing equally spaced sampling on the fused line segments based on a predetermined sampling interval; or - performing equally spaced interval sampling on each edge straight line segment included in the fused line segment based on a predetermined sampling interval; or - Interval sampling is performed based on the length of each edge straight line segment included in the fused line segment.

6. The method according to claim 5, characterized in that In performing the interval sampling based on the length of each edge straight line segment included in the fused line segment, the number and / or interval of sampling points for each edge straight line segment is determined according to a piecewise function on the length.

7. The method according to any one of claims 1 to 3, characterized in that Before implementing edge straight line segment detection, a convolutional neural network binary classifier trained with an image of a specified area containing at least a complete strip of decorative parts as a positive sample is used to determine whether the acquired image contains a valid image of the specified area of ​​the strip of decorative parts.

8. The method according to claim 7, characterized in that The binary classifier uses the MobileNet neural network.

9. The method according to any one of claims 1 to 3, characterized in that The strip-shaped decorative piece is a light strip, a metal side strip, a colored side strip or a wooden side strip. 10 . A computer-readable storage medium having executable instructions, which, when executed, cause a computer to perform the method according to claim 1 .

11. A device for identifying a strip-shaped trim of a vehicle, the device comprising: - a 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 one of claims 1 to 9.

12. A system for identifying a strip trim of a vehicle, the system comprising: - an industrial camera for capturing an image of a defined area of ​​the strip; - Device for identifying a strip of trim on a vehicle according to claim 11; - an alarm device for outputting an alarm signal when it is determined that the belt-shaped fastener is not present.

13. The system according to claim 12, characterized in that The industrial camera is arranged at a location without human intervention.

14. The system according to claim 12 or 13, characterized in that The system comprises a data interface configured to: Obtaining information about the trim strip from the vehicle manufacturing system; and / or The strip trim recognition result is sent to the vehicle manufacturing system.

15. Vehicle production line comprising a system for identifying a trim strip of a vehicle according to any one of claims 12 to 14.

Images