A vehicle-mounted display screen alignment and fitting feature searching method and computer equipment

Abstract

The application discloses a kind of vehicle-mounted display screen alignment and fit feature search method, comprising: S1: using camera to obtain the original image of the object to be searched for feature;S2: the continuous curve function f (x, y) projected on plane of the feature to be searched for obtained;S3: average take n points on curve function f (x, y), generate n rectangle with n points as center, height h and width w;S4: rotate rectangle to obtain the n rectangle roi region corresponding to original image;S5: each rectangle roi region is carried out convolution operation;S6: the maximum value of intensity S is obtained as the strongest edge point on rectangle roi region;S7: the coordinates of the strongest edge point on rectangle roi region are restored to original image coordinate system;S8: n original-strongest edge points are screened out the points on the same curve, to obtain final feature curve.The application can improve feature extraction speed, the accuracy and robustness of feature search, improve production efficiency and yield.

Description

Technical Field

[0001] This invention relates to the field of image recognition technology, and more specifically to a method and computer device for finding alignment and bonding features of an in-vehicle display screen. Background Technology

[0002] With the increasing prevalence of automobiles, in-vehicle displays are offering more and more functions, placing higher demands on their manufacturing processes. One crucial process is alignment and bonding, which involves attaching the display module to the back cover.

[0003] Because the alignment and bonding process requires high precision, the existing technology mainly uses visual-assisted positioning. However, in the actual production process, due to the different sizes and colors of the incoming materials, there are defects or foreign objects, which makes the product features unclear or partially damaged. For example, when the common product features are changed from straight lines to curves, the product features cannot be accurately located, and the yield rate of production cannot be guaranteed. Summary of the Invention

[0004] The present invention provides a method and computer device for finding alignment and bonding features of vehicle display screens that effectively improves the feature extraction speed, which can at least solve one of the above-mentioned technical problems.

[0005] To solve the above-mentioned technical problems, the present invention adopts the following technical solution:

[0006] A method for finding alignment and bonding features of an in-vehicle display screen includes the following steps:

[0007] S1: Use a camera to acquire the original image of the object whose features are to be found;

[0008] S2: Based on the continuous curve function f(x, y) projected onto the plane from the drawing of the object to be searched or the actual obtained feature to be searched, denote the starting point of the part to be searched on the curve function f(x, y) as P. s The endpoint is P. e ;

[0009] S3: First, based on the preset quantity n, P on the curve function f(x, y) s To P e Within the interval, n points are selected on average according to the curve length. Then, based on the preset height h and width w, n rectangles r1, r2, ..., r' are generated, each centered at one of these n points with a height h and a width w. n ;

[0010] S4: Rotate rectangles r1, r2, ..., r3 respectively. nContinue this process until the height of each rectangle is parallel to the y-axis and the width is parallel to the x-axis, resulting in n rectangular roi regions R1, R2, ..., Rn corresponding to the original image. n ;

[0011] S5: Using a vertical edge convolution kernel of a set size and type, perform convolution operation on each rectangular ROI region to obtain the edge intensity in the y direction of each point;

[0012] S6: Let Str be the edge strength value in the y-direction of each point, and Dis be the distance between each point and the preset reference point. Multiply both by their respective weights W. str and W dis Then, the sums are added together, and the maximum value of the resulting intensity S is taken as the strongest edge point P on the rectangular roi region. n ;

[0013] S7: P the strongest edge points n The coordinates of each rectangular ROI region are restored to the original image coordinate system, resulting in n original-strongest edge points p1, p2, ..., p... n ;

[0014] S8: Using the RANSAC random sample consensus algorithm, select the points on the same curve from the n original strongest edge points to obtain the final feature curve.

[0015] Furthermore, in S1, several light sources are deployed around the object to help highlight its features.

[0016] Furthermore, in S2, the drawing of the object to be feature-searched refers to a 2D drawing.

[0017] Furthermore, in S3, each rectangle r1, r2, ..., r n The width direction is parallel to the tangent direction of the curve function f(x, y) at the corresponding center point.

[0018] Furthermore, in S5, the convolution kernel is either a Prewitt convolution kernel or a Sobel convolution kernel.

[0019] Furthermore, in S6, the formula for calculating the intensity S is: S = Str * W str +Dis*W dis .

[0020] Furthermore, S7 further includes:

[0021] S71: Let R be the rectangular region roi. n The center point is C n , build with C n Starting from Pn Vector V with endpoint n ;

[0022] S72: For the rectangular roi region R n The rectangle before rotation is r. n Let rectangle r be... n The center point is c;

[0023] S73: Transfer vector V n Rotation, by R n Rotate to r n From the angle, we obtain a new vector v n ;

[0024] S74: Transfer the new vector v n The starting point is set to rectangle r. n center point c n Then the new vector v n The endpoint is the rectangular roi region R. n The strongest edge point P on n The original strongest edge point p, restored to the original image coordinate system n .

[0025] Furthermore, S8 further includes:

[0026] S81: Randomly select the minimum number of points from n original-strongest edge points that can form the curve function f(x,y) after translation, rotation and scaling. The number of these points is m. Generate the curve function g(x,y) based on these m points, so that the curve function g(x,y) can be obtained by translation, rotation and scaling only.

[0027] S82: Calculate the distance from each of the remaining nm original-strongest edge points to the curve function g(x, y): When the distance is not greater than a preset value, record the point as a point on the curve and add it to the point set M; when the distance is greater than the preset value, record the point as a point not on the curve and add it to the point set M'. Also, add the m points used in S81 to the point set M, and obtain the size n of the point set M. M ;

[0028] S83: Repeat steps S81-S82 until the preset number of iterations is reached, and obtain n from these iterations. M The set of points M corresponding to the maximum value MAX and M' MAX ;

[0029] S84: Use the minimum error fitting method to fit the point set M MAXThe curve function f'(x, y) is fitted to the points in the graph so that the curve function f'(x, y) can be obtained by translation, rotation and scaling.

[0030] S85: The curve function f'(x, y) is the final feature curve to be found.

[0031] Furthermore, in step S83, when repeating steps S81-S82, a random iteration method or a permutation and combination method can be used for cyclical processing.

[0032] A computer device includes a memory and a processor, the memory storing a computer program that, when executed by the processor, causes the processor to perform the steps of the vehicle display screen alignment feature lookup method.

[0033] The beneficial effects of this invention are reflected in:

[0034] This invention aims to utilize visual-assisted positioning technology to accurately locate product features even when they undergo occasional changes or are partially damaged. In particular, it can still accurately identify and locate product features when they change from straight lines to curves. This not only improves the speed of feature extraction but also enhances the accuracy and robustness of feature lookup, thereby increasing production efficiency and yield. Attached Figure Description

[0035] The accompanying drawings, which are provided to further illustrate this application and form part of this application, illustrate exemplary embodiments of this application and are used to explain this application, but do not constitute an undue limitation of this application.

[0036] Figure 1 This is a schematic diagram of the overall process of the feature search method according to an embodiment of the present invention.

[0037] Figure 2 These are n rectangles r1, r2, ... and r in this embodiment of the invention. n A schematic diagram on the curve function f(x, y).

[0038] Figure 3 These are the original images acquired by the camera in this embodiment of the invention.

[0039] Figure 4 This is a schematic diagram of the n rectangles generated in an embodiment of the present invention.

[0040] Figure 5 The rectangle roi is obtained by rotating one of the rectangles in an embodiment of the present invention.

[0041] Figure 6 This is a schematic diagram of the edge strength obtained after convolution in an embodiment of the present invention.

[0042] Figure 7 This is a schematic diagram showing the distance between a point and a preset reference point in an embodiment of the present invention.

[0043] Figure 8 This is a schematic diagram of the intensity S according to an embodiment of the present invention.

[0044] Figure 9 This is a schematic diagram of the point where the intensity S is at its maximum according to an embodiment of the present invention.

[0045] Figure 10 This is a schematic diagram of the original-strongest edge points restored to the original image according to an embodiment of the present invention.

[0046] Figure 11 This is a schematic diagram of the final feature curve obtained by fitting according to an embodiment of the present invention.

[0047] Figure 12 This is a demonstration diagram of the RANSAC random sample consensus algorithm according to an embodiment of the present invention.

[0048] Figure 13 This is a structural block diagram of a computer device according to an embodiment of the present invention. Detailed Implementation

[0049] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0050] It should be noted that the meaning of "and / or" throughout the text includes three parallel solutions. Taking "A and / or B" as an example, it includes solution A, solution B, or a solution where both A and B are satisfied. Additionally, "multiple" refers to two or more. Furthermore, those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0051] See Figure 1 This invention provides a method for finding alignment and bonding features of an in-vehicle display screen, comprising the following steps:

[0052] S1: Use a camera to acquire the original image of the object whose features are to be found;

[0053] S2: Based on the continuous curve function f(x, y) projected onto the plane from the drawing of the object to be searched or the actual obtained feature to be searched, denote the starting point of the part to be searched on the curve function f(x, y) as P. s The endpoint is P. e ;

[0054] S3: First, based on the preset quantity n, P on the curve function f(x, y) s To P e Within the interval, n points are selected on average according to the curve length. Then, based on the preset height h and width w, n rectangles r1, r2, ..., r' are generated, each centered at one of these n points with a height h and a width w. n ;

[0055] S4: Rotate rectangles r1, r2, ..., r3 respectively. n Continue this process until the height of each rectangle is parallel to the y-axis and the width is parallel to the x-axis, resulting in n rectangular roi regions R1, R2, ..., Rn corresponding to the original image. n ;

[0056] S5: Using a vertical edge convolution kernel of a set size and type, perform convolution operation on each rectangular ROI region to obtain the edge intensity in the y direction of each point;

[0057] S6: Let Str be the edge strength value in the y-direction of each point, and Dis be the distance between each point and the preset reference point. Multiply both by their respective weights W. str and W dis Then, the sums are added together, and the maximum value of the resulting intensity S is taken as the strongest edge point P on the rectangular roi region. n ;

[0058] S7: P the strongest edge points n The coordinates of each rectangular ROI region are restored to the original image coordinate system, resulting in n original-strongest edge points p1, p2, ..., p... n ;

[0059] S8: Using the RANSAC random sample consensus algorithm, select the points on the same curve from the n original strongest edge points to obtain the final feature curve.

[0060] This invention aims to utilize visual-assisted positioning technology to accurately locate product features even when they undergo occasional changes or are partially damaged. In particular, it can still accurately identify and locate product features when they change from straight lines to curves. This not only improves the speed of feature extraction but also enhances the accuracy and robustness of feature lookup, thereby increasing production efficiency and yield.

[0061] See Figure 1 and 3 In this embodiment, in step S1, several light sources are deployed around the object to help highlight its features, so as to obtain the features of the object better, clearer and faster. It should be noted that the object mentioned in this application refers to the vehicle display screen.

[0062] See Figure 1 and 3 In this embodiment, in step S2, the drawing of the object to be feature-finded refers to a 2D drawing. 2D drawings provide a more precise description of shapes, especially curves, to indicate the processing method. While 3D graphics can also be used to obtain feature curves through 3D software, in practical applications, if the curve shape is complex, there may be slight errors due to differences in the performance of 3D software, which could affect the accuracy of feature finding.

[0063] See Figure 2 and 4 In this embodiment, in S3, each rectangle r1, r2, ..., r n The width direction of each of the n points is parallel to the tangent direction of the curve function f(x, y) at the corresponding center point. s To P e Distributed evenly within the interval, taking each point as the center, n rectangles of equal size r1, r2, ..., r3 with height h and width w are selected. n Let the rectangles be r1, r2, ..., r n The extension lines along the width direction are a1, a2, ..., a n Let b1, b2, ..., bn be the extensions of the tangent lines of n points on the curve function f(x, y). n a1, a2, ... and a n respectively with b1, b2, ... and b n parallel.

[0064] See Figure 6 In this embodiment, in step S5, the convolution kernel can be either a Prewitt convolution kernel or a Sobel convolution kernel.

[0065] For example:

[0066] A 3x3 Prewitt convolution kernel is:

[0067] A 3x3 Sobel convolution kernel is:

[0068] Compared to the Prewitt convolutional kernel, the Sobel convolutional kernel is more robust to noise due to the addition of edge weights.

[0069] The edge strength obtained after convolution is as follows Figure 6 As shown, the brighter the edge, the greater its intensity.

[0070] See Figure 7 and 8 In this embodiment, in step S6, the formula for calculating the intensity S is: S = Str * W str +Dis*W dis .

[0071] The preset reference point is a pre-defined coordinate point fixed within a rectangular ROI, used to control the position of the strongest edge point found earlier on the ROI. In practice, this coordinate point is typically set differently based on specific features or experience, and its distance from the preset reference point is as follows: Figure 7 As shown, the brighter the light, the closer the distance.

[0072] The intensity S obtained by weighting and summing the edge intensity and distance is as follows: Figure 8 As shown, the brighter the light, the greater the intensity.

[0073] See Figure 5 , 9 In this embodiment, S7 further includes:

[0074] S71: Let R be the rectangular region roi. n The center point is C n , build with C n Starting from P n Vector V with endpoint n ;

[0075] S72: For the rectangular roi region R n The rectangle before rotation is r. n Let rectangle r be... n The center point is c;

[0076] S73: Transfer vector V n Rotation, by R n Rotate to r n From the angle, we obtain a new vector v n ;

[0077] S74: Transfer the new vector v n The starting point is set to rectangle r. n center point c n Then the new vector v n The endpoint is the rectangular roi region R. n The strongest edge point P on n The original strongest edge point p, restored to the original image coordinate system n .

[0078] Will Figure 4 The rectangle ROI obtained by rotating a rectangle is as follows: Figure 5 As shown, according to Figure 8 The point where the intensity S is maximized is found as follows: Figure 9 As shown, the original strongest edge points restored to the original image are as follows: Figure 10 As shown.

[0079] See Figure 11 and 12 In this embodiment, S8 further includes:

[0080] S81: Randomly select the minimum number of points from n original-strongest edge points that can form the curve function f(x,y) after translation, rotation and scaling. The number of these points is m. Generate the curve function g(x,y) based on these m points, so that the curve function g(x,y) can be obtained by translation, rotation and scaling only.

[0081] S82: Calculate the distance from each of the remaining nm original-strongest edge points to the curve function g(x, y): When the distance is not greater than a preset value, record the point as a point on the curve and add it to the point set M; when the distance is greater than the preset value, record the point as a point not on the curve and add it to the point set M'. Also, add the m points used in S81 to the point set M, and obtain the size n of the point set M. M ;

[0082] S83: Repeat steps S81-S82 until the preset number of iterations is reached, and obtain n from these iterations. M The set of points M corresponding to the maximum value MAX and M' MAX ;

[0083] S84: Use the minimum error fitting method to fit the point set M MAX The curve function f'(x, y) is fitted to the points in the graph so that the curve function f'(x, y) can be obtained by translation, rotation and scaling.

[0084] S85: The curve function f'(x, y) is the final feature curve to be found.

[0085] The final feature curve of the fit is shown below. Figure 11 As shown in the figure, the demonstration effect of the RANSAC random sample consensus algorithm is as follows. Figure 12 As shown.

[0086] See Figure 1 In this embodiment, during step S83, when repeating steps S81-S82, a random iteration method or a permutation and combination method can be used for looping. This embodiment shows the random iteration method. In addition, a permutation and combination method can also be used to ensure that each combination of points is calculated once, thereby improving the accuracy of feature search results.

[0087] See Figure 13 This invention also provides a computer device, including a memory and a processor. The memory stores a computer program, which, when executed by the processor, causes the processor to perform the steps of the vehicle display screen alignment and bonding feature lookup method. The steps of the vehicle display screen alignment and bonding feature lookup method described above can be the steps in the vehicle display screen alignment and bonding feature lookup methods of the various embodiments above.

[0088] In summary, this invention aims to utilize visual-assisted positioning technology to accurately locate product features even when they undergo occasional changes or are partially damaged. In particular, it can still accurately identify and locate product features when they change from straight lines to curves. This not only improves the speed of feature extraction but also enhances the accuracy and robustness of feature lookup, thereby increasing production efficiency and yield.

[0089] It should be understood that the examples and embodiments described herein are for illustrative purposes only and are not intended to limit the invention. Those skilled in the art can make various modifications or changes based on them. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the invention should be included within the protection scope of the invention.

Claims

1. A method for finding alignment and bonding features of a vehicle-mounted display screen, characterized in that, Includes the following steps: S1: Use a camera to acquire the original image of the object whose features are to be found; S2: Based on the continuous curve function f(x, y) projected onto the plane from the drawing of the object to be searched or the actual obtained feature to be searched, denote the starting point of the part to be searched on the curve function f(x, y) as P. s The endpoint is P. e ; S3: First, based on the preset quantity n, P on the curve function f(x, y) s To P e Within the interval, n points are selected on average according to the curve length. Then, based on the preset height h and width w, n rectangles r1, r2, ..., r' are generated, each centered at one of these n points with a height h and a width w. n ; S4: Rotate rectangles r1, r2, ..., r3 respectively. n Continue this process until the height of each rectangle is parallel to the y-axis and the width is parallel to the x-axis, resulting in n rectangular roi regions R1, R2, ..., Rn corresponding to the original image. n ; S5: Using a vertical edge convolution kernel of a set size and type, perform convolution operation on each rectangular ROI region to obtain the edge intensity in the y direction of each point; S6: Let Str be the edge strength value in the y-direction of each point, and Dis be the distance between each point and the preset reference point. Multiply both by their respective weights W. str and W dis Then, the sums are added together, and the maximum value of the resulting intensity S is taken as the strongest edge point P on the rectangular roi region. n ; S7: P the strongest edge points n The coordinates of each rectangular ROI region are restored to the original image coordinate system, resulting in n original-strongest edge points p1, p2, ..., p... n ; S8: Using the RANSAC random sample consensus algorithm, select the points on the same curve from the n original strongest edge points to obtain the final feature curve.

2. The method for finding alignment and bonding features of a vehicle-mounted display screen as described in claim 1, characterized in that, In step S1, several light sources are deployed around the object to help highlight its features.

3. The method for finding alignment and bonding features of a vehicle-mounted display screen as described in claim 1, characterized in that, In S2, the drawing of the object to be feature-searched refers to a 2D drawing.

4. The method for finding alignment and bonding features of a vehicle-mounted display screen as described in claim 1, characterized in that, In S3, each rectangle r1, r2, ..., r n The width direction is parallel to the tangent direction of the curve function f(x, y) at the corresponding center point.

5. The method for finding alignment and bonding features of a vehicle-mounted display screen as described in claim 1, characterized in that, In S5, the convolution kernel is either a Prewitt convolution kernel or a Sobel convolution kernel.

6. The method for finding alignment and bonding features of a vehicle-mounted display screen as described in claim 1, characterized in that, In S6, the formula for calculating the strength S is: S = Str * W str +Dis*W dis .

7. The method for finding alignment and bonding features of a vehicle-mounted display screen as described in claim 1, characterized in that, S7 further includes: S71: Let R be the rectangular region roi. n The center point is C n , build with C n Starting from P n Vector V with endpoint n ; S72: For the rectangular roi region R n The rectangle before rotation is r. n Let rectangle r be... n The center point is c; S73: Transfer vector V n Rotation, by R n Rotate to r n From the angle, we obtain a new vector v n ; S74: Transfer the new vector v n The starting point is set to rectangle r. n center point c n Then the new vector v n The endpoint is the rectangular roi region R. n The strongest edge point P on n The original strongest edge point p, restored to the original image coordinate system n .

8. The method for finding alignment and bonding features of a vehicle-mounted display screen as described in claim 1, characterized in that, S8 further includes: S81: Randomly select the minimum number of points from n original-strongest edge points that can form the curve function f(x,y) after translation, rotation and scaling. The number of these points is m. Generate the curve function g(x,y) based on these m points, so that the curve function g(x,y) can be obtained by translation, rotation and scaling only. S82: Calculate the distance from each of the remaining nm original-strongest edge points to the curve function g(x, y): When the distance is not greater than a preset value, record the point as a point on the curve and add it to the point set M; when the distance is greater than the preset value, record the point as a point not on the curve and add it to the point set M'. Also, add the m points used in S81 to the point set M, and obtain the size n of the point set M. M ; S83: Repeat steps S81-S82 until the preset number of iterations is reached, and obtain n from these iterations. M The set of points M corresponding to the maximum value MAX and M' MAX ; S84: Use the minimum error fitting method to fit the point set M MAX The curve function f'(x, y) is fitted to the points in the graph so that the curve function f'(x, y) can be obtained by translation, rotation and scaling. S85: The curve function f'(x, y) is the final feature curve to be found.

9. The method for finding alignment and bonding features of a vehicle-mounted display screen as described in claim 8, characterized in that, In step S83, when repeating steps S81-S82, a random iteration method or a permutation and combination method can be used for cyclical processing.

10. A computer device comprising a memory and a processor, the memory storing a computer program that, when executed by the processor, causes the processor to perform the steps of the method as claimed in any one of claims 1-9.

Images