A method, system, and terminal for extracting the center of a line laser stripe based on EAST.

Abstract

This invention belongs to the field of information technology service technology and discloses a method, system, and terminal for extracting the center of a linear laser stripe based on EAST. First, a calibrated camera is used to acquire laser stripe images. Median filtering is used to denoise the images and remove isolated noise points. An EAST network is trained using manually labeled images to identify the light stripe target and separate the light stripe from the background. Then, the sub-pixel coordinates of the light stripe are extracted using the traditional gray-scale centroid method. A cubic B-spline curve is used to fit the initial center line, effectively solving the problem of discontinuities in the center line of some light stripes. Finally, Savitzky-Golay filtering is used to optimize the center line, achieving accurate extraction of the light stripe center line. This invention, based on the traditional gray-scale centroid method, introduces cubic B-spline curve fitting to reconnect discontinuous center lines, ensuring the accuracy and real-time performance of the light stripe center extraction. Furthermore, this method exhibits good characteristics in terms of extraction accuracy, processing speed, and noise resistance.

Description

Technical Field

[0001] This invention belongs to the field of information technology service technology, and in particular relates to a method, system and terminal for extracting the center of a line laser stripe based on EAST. Background Technology

[0002] 3D scanning reconstruction based on line structured light is widely used due to its low cost, simple operation, high speed, and high accuracy, especially in industrial production, topographic exploration, and underwater resource detection. This technology uses a calibrated camera to acquire laser stripe images in real time and transmits them to a PC for image processing, ultimately reconstructing the 3D point cloud information of the target object. Accurately and efficiently extracting the center of the laser stripes is a key factor affecting the system's accuracy and real-time performance.

[0003] Centering extraction mainly includes two steps: image preprocessing and extraction of sub-pixel coordinates of the light stripe center. The most commonly used traditional centering extraction algorithms are the Steger method and the gray-scale centroid method. The Steger method has high accuracy, but it requires a large number of convolution operations, resulting in high computational complexity and making real-time extraction impossible. The gray-scale centroid method, on the other hand, is fast, but it is susceptible to noise interference, which can lead to a decrease in extraction accuracy.

[0004] For target objects with complex surfaces, the image quality is low due to environmental noise interference and uneven lighting, and the obtained center line may be broken or have outliers, making it impossible to accurately extract the center line. In this process, image segmentation inevitably introduces some errors.

[0005] Based on the above analysis, the problems and shortcomings of the existing technology are as follows:

[0006] 1) High computational complexity: Traditional center extraction algorithms, such as the Steger method, although highly accurate, have high computational complexity due to the large number of convolution operations required, making it difficult to achieve real-time extraction.

[0007] 2) Susceptible to noise interference: While the grayscale centroid method offers fast extraction speed, it is easily affected by noise, leading to a decrease in extraction accuracy. In actual working environments, factors such as lighting conditions, environmental noise, and surface materials may influence the extraction results.

[0008] 3) Image quality issues: For target objects with complex surfaces, environmental noise interference and uneven surface illumination may lead to low image quality, thus affecting the accuracy of centerline extraction.

[0009] 4) Centerline breakage and outliers: The centerline of the laser stripe on a complex surface target object may be broken or exhibit outliers, which will affect the accuracy of the three-dimensional reconstruction of the target object.

[0010] 5) Image segmentation error: Image segmentation may introduce certain errors during the extraction of the center line of the laser stripe, thus affecting the final extraction accuracy. Summary of the Invention

[0011] To address the problems existing in the prior art, this invention provides a method, system, and terminal for extracting the center of line laser stripes based on EAST. This solves the problems of discontinuity and outliers in the center of the light stripes, which are difficult to extract, as well as the accuracy issues in the prior art, thereby improving the accuracy and real-time performance of the 3D scanning and reconstruction system.

[0012] This invention is implemented as follows: a method for extracting the center of a line laser stripe based on EAST. The method includes: first, acquiring laser stripe images using a calibrated camera; performing noise reduction processing on the image using median filtering to remove isolated noise points; training an EAST network using manually labeled images to identify the light stripe target and achieve separation of the light stripe from the background; then, extracting the sub-pixel coordinates of the light stripe using the traditional gray-scale centroid method; fitting the initial centerline using a cubic B-spline curve to effectively solve the problem of discontinuities in the centerline of some light stripes; and finally, optimizing the centerline using Savitzky-Golay filtering to achieve accurate extraction of the light stripe centerline.

[0013] Furthermore, the overall framework for image segmentation and detection includes: employing a scene text detection algorithm, namely the EAST detection algorithm; designing a deep neural network; manually annotating the effective region of the stripes and the original laser image; and using these annotations as inputs to train the EAST deep learning model; using the trained model to detect the light stripe image under test, directly predicting the light stripe region in the overall image; then using the non-maximum suppression (NMS) method to filter out redundant bounding boxes; and simultaneously employing the backpropagation algorithm to quickly converge the network to obtain a better training model, thereby obtaining the final result. Finally, the output is the coordinates of the four coordinate points of the light stripe bounding box.

[0014] Furthermore, the central extraction includes:

[0015] Step 1: Extract the initial center line using the grayscale centroid method;

[0016] Step 2: Reconnect the broken lines using cubic B-spline curve fitting;

[0017] Step 3: The Savitzky-Golay filtering algorithm is used to smooth and denoise the center line.

[0018] Furthermore, in step one, the grayscale centroid method involves sorting the grayscale values ​​of the light stripe within the region of interest, extracting the centroid of the grayscale values ​​of each row (column) of each pixel in the cross-section of the light stripe, and using the grayscale values ​​as weights to represent the centerline of the light stripe. The calculation formula is as follows:

[0019]

[0020] In the formula (x i ,y i F(x) represents the coordinates of a pixel in a row (column) of the light stripe. i ,y i F(x) represents the grayscale value at the corresponding coordinates, M represents the number of pixels on the cross-section of the light stripe, and F(x) represents the number of pixels on the cross-section of the light stripe. i ,y i ) represents the coordinates of the center point of the light stripe obtained from this cross section.

[0021] Furthermore, in step two, the formula for the B-spline function is as follows:

[0022]

[0023] Where C i (i = 0, 1, ..., n) are the control vertices, N i,k (t)(i=1,2...,n) is a k-order B-spline basis function;

[0024] Where the basis function N i,k (t) is:

[0025]

[0026] The equation of the cubic B-spline curve is:

[0027] G(t) = G0 * N 0,3 (t)+G1*N 1,3 (t)+G2*N 2,3 (t)+G3*N 3,3 (t);

[0028] 1) First, n type value points Q are given. i (i = 0, 1, ... n-1), that is, there are n-1 cubic B-spline curves;

[0029] 2) Based on the given shape points, find the vertices of the corresponding B-spline curve, shape points {Q}. i} and control point {C i The relationship between} is (C i-1 +4C i +C i+1 ) / 6=Q i(i = 0, 1, ... n-1), this system of equations has n equations and n+2 unknowns, so the boundary conditions at the endpoints need to be added;

[0030] 3) Determine the boundary conditions for the cubic spline curve. For non-closed curves, take...

[0031]

[0032] By combining the above n equations, the control point C can be determined. i The position coordinates of (i = 0, 1, ..., n-1);

[0033] 4) Constructing the interpolation matrix: Construct the interpolation matrix for the cubic B-spline curve. The interpolation matrix is ​​a 4x4 matrix, calculated based on the basis functions of the B-spline curve;

[0034] 5) Construct an interpolation curve using the interpolation matrix and control points, ultimately obtaining a curve passing through n type value points Q. i Cubic B-spline curves for (i = 0, 1, ..., n-1);

[0035] 6) Optimize and adjust the fitted curve as needed, by adjusting the node positions and weight allocation, to improve the curve fitting results.

[0036] Furthermore, step three includes: during the filtering process, the filter window width and the order of the polynomial fitting determine the filtering effect. The specific steps are as follows:

[0037] 1) Let the sliding window size be 2M+1, and the original data points in the window be x(n), where n takes the value [-M,...,0,...,M]. Construct a polynomial of order N for fitting:

[0038]

[0039] In the formula, k is the order of the polynomial, and d k The coefficients of the polynomial;

[0040] 2) Least squares fitting residuals

[0041]

[0042] 3) To minimize E, let the derivative of E with respect to each coefficient be 0, i.e.

[0043]

[0044] The polynomial coefficients d can be obtained. k ;

[0045] 4) Each time the sliding window is moved by 2M+1, the value of this polynomial at the center point of the window is the filtered result. When the x-coordinate is 0, the filtered value is f(0)=d0.

[0046] Another object of the present invention is to provide an EAST-based line laser stripe center extraction system that applies the aforementioned EAST-based line laser stripe center extraction method. The EAST-based line laser stripe center extraction system includes:

[0047] The image acquisition module is used to acquire images of laser stripes.

[0048] The image preprocessing module is used for median filtering and noise reduction, and for ROI region extraction using the EAST deep learning model.

[0049] The centerline extraction module is used to extract the initial centerline using the grayscale centroid method;

[0050] The centerline fitting module is used to fit the initial centerline using cubic B-splines;

[0051] The centerline optimization module is used to optimize the centerline using Savitzky-Golay filtering.

[0052] Another object of the present invention is to provide a computer device, the computer device including a memory and a processor, the memory storing a computer program, which, when executed by the processor, causes the processor to perform the steps of the EAST-based line laser stripe center extraction method.

[0053] Another object of the present invention is to provide a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the steps of the EAST-based line laser stripe center extraction method.

[0054] Another objective of this invention is to provide an information data processing terminal for implementing the EAST-based line laser stripe center extraction system.

[0055] Based on the above technical solutions and the technical problems solved, the advantages and positive effects of the technical solution to be protected by this invention are as follows:

[0056] First, this invention effectively removes interference from environmental noise and uneven light reflection on light stripe extraction by constructing a deep learning-based laser stripe segmentation algorithm, achieving accurate separation of the light stripe region and the background. Furthermore, based on the traditional grayscale centroid method, it introduces cubic B-spline curve fitting to reconnect discontinuous center lines, and combines this with the Savitzky-Golay filtering algorithm to smooth and denoise the center lines, ensuring the accuracy and real-time performance of light stripe center extraction. Simultaneously, this invention exhibits superior characteristics in terms of extraction accuracy, processing speed, and noise resistance.

[0057] Second, considering the technical solution as a whole or from a product perspective, the technical effects and advantages of the technical solution to be protected by this invention are specifically described as follows:

[0058] The standard error of the algorithm in this invention is the smallest compared to the other two algorithms, therefore the algorithm in this invention has the highest accuracy. The Steger algorithm has the largest standard error of light stripes, and the accuracy of the traditional gray-scale centroid method is also slightly lower than that of the algorithm in this invention.

[0059] The traditional grayscale centroid method has an average processing time of approximately 0.053 seconds, while the proposed method has an average processing time of 0.050 seconds. Therefore, the time difference between the proposed method and the traditional grayscale centroid method is not significant. In contrast, the Steger algorithm has an average processing time of approximately 0.268 seconds, meaning the algorithm in this invention improves processing time by 20 times compared to the Steger algorithm. However, the Steger method requires a large number of convolution operations, resulting in a high computational cost and low efficiency in extracting the center line.

[0060] Third, as supplementary evidence of the inventive step of the claims of this invention, it is also reflected in the following important aspects:

[0061] High precision: The EAST-based method for extracting the center of a line laser stripe improves the accuracy of separating the light stripe from the background by training the EAST network to identify the laser stripe image, thereby improving the precision of center line extraction.

[0062] Real-time performance: The training and prediction process of the EAST network is relatively fast, enabling real-time extraction of the light stripe centerline.

[0063] Noise resistance: Median filtering is used for noise reduction, which can effectively remove isolated noise points and reduce the impact of noise on centerline extraction.

[0064] Continuity: The initial centerline is fitted with a cubic B-spline curve, which effectively solves the problem of discontinuity in the centerline of some light stripes.

[0065] Optimization: Savitzky-Golay filtering optimizes the centerline, further improving the extraction accuracy of the light stripe centerline. Attached Figure Description

[0066] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the embodiments of the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0067] Figure 1 This is a flowchart of the overall process for extracting the center of a line structured light according to an embodiment of the present invention;

[0068] Figure 2 This is a flowchart of the effective region selection process for light stripes based on EAST, provided in an embodiment of the present invention.

[0069] Figure 3 This is a laser image segmentation effect diagram provided in an embodiment of the present invention;

[0070] Figure 4 These are simulation results of the algorithm provided in this embodiment of the invention; wherein, (a) is the result of extracting the center line using the gray-scale centroid method; (b) is a magnified view of a portion of (a); (c) is the result of fitting the center line after cubic B-spline curve; (d) is a magnified view of a portion of (c); (e) is the result after optimization using Savitzky-Golay filtering; and (f) is a magnified view of the final result.

[0071] Figure 5 These are simulation results of different stripe centerline extraction provided in the embodiments of the present invention; wherein (a) is complex; (b) has a small curvature; (c) has a large curvature; and (d) is general. Detailed Implementation

[0072] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0073] To address the problems existing in the prior art, the present invention provides a method, system, and terminal for extracting the center of a line laser stripe based on EAST. The present invention will be described in detail below with reference to the accompanying drawings.

[0074] like Figure 1 As shown, the EAST-based line laser stripe center extraction method provided in this embodiment of the invention includes:

[0075] First, laser stripe images are acquired using a calibrated camera. Median filtering is then used to reduce noise and remove isolated noise points. The EAST network is trained using manually labeled images to identify the light stripe targets and separate the light stripes from the background. Next, the sub-pixel coordinates of the light stripes are extracted using the traditional grayscale centroid method. A cubic B-spline curve is then used to fit the initial centerline, effectively solving the problem of discontinuities in the centerline of some light stripes. Finally, Savitzky-Golay filtering is used to optimize the centerline, achieving accurate extraction of the light stripe centerline.

[0076] The above method can effectively avoid the influence of environmental factors such as noise, uneven lighting and random occlusion of light stripes during the extraction of the center of the line structured light stripe, and ensure the robustness, accuracy and real-time performance of the subsequent extraction.

[0077] like Figure 2 As shown, the overall framework for image segmentation and detection provided in this embodiment of the invention includes:

[0078] A scene text detection algorithm, namely the EAST detection algorithm, is adopted. A deep neural network is designed, and the effective region of the stripe and the original laser image are manually labeled and used as input to train the EAST deep learning model. The trained model is used to detect the light stripe image to be tested, directly predicting the light stripe region in the overall image. Then, the non-maximum suppression (NMS) method is used to filter out redundant bounding boxes. At the same time, the backpropagation algorithm is used to quickly converge the network to obtain a better training model, thus obtaining the final result. Finally, the output is the coordinates of the four coordinate points of the light stripe bounding box.

[0079] 3. Central Extraction

[0080] a. First, the initial center line is extracted using the gray-scale centroid method. The gray-scale centroid method sorts the gray values ​​of the light stripe within the region of interest, extracts the centroid of the gray values ​​for each row (column) of each pixel in the cross-section of the light stripe, and uses these gray values ​​as weights to represent the center line of the light stripe. The calculation formula is as follows:

[0081]

[0082] In the formula (x i ,y i F(x) represents the coordinates of a pixel in a row (column) of the light stripe. i ,y i F(x) represents the grayscale value at the corresponding coordinates, M represents the number of pixels on the cross-section of the light stripe, and F(x) represents the number of pixels on the cross-section of the light stripe. i ,y i ) represents the coordinates of the center point of the light stripe obtained from this cross section.

[0083] b. To address the issue of discontinuities in the centerline extracted by the traditional gray-scale centroid method, cubic B-spline curve fitting is used to reconnect broken lines.

[0084] The formula for the B-spline function is as follows:

[0085]

[0086] Where C i (i = 0, 1, ..., n) are the control vertices, N i,k (t)(i=1,2...,n) is a k-order B-spline basis function.

[0087] Where the basis function N i,k (t) is:

[0088]

[0089]

[0090]

[0091]

[0092] The equation of the cubic B-spline curve is:

[0093] G(t) = G0 * N 0,3 (t)+G1*N 1,3 (t)+G2*N 2,3 (t)+G3*N 3,3 (t)

[0094] 1) First, n type value points Q are given. i (i = 0, 1, ... n-1), that is, there are n-1 cubic B-spline curves.

[0095] 2) Based on the given shape points, find the vertices of the corresponding B-spline curve, shape points {Q}. i} and control point {C i The relationship between} is (C i-1 +4C i +C i+1 ) / 6=Q i (i = 0, 1, ..., n-1). This system of equations has n equations and n+2 unknowns, therefore, boundary conditions at the endpoints need to be added.

[0096] 3) Determine the boundary conditions for the cubic spline curve. For non-closed curves, take...

[0097]

[0098] By combining the above n equations, the control point C can be determined.i The position coordinates of (i = 0, 1, ..., n-1).

[0099] 4) Constructing the interpolation matrix: Construct the interpolation matrix for the cubic B-spline curve. The interpolation matrix is ​​a 4x4 matrix, calculated based on the basis functions of the B-spline curve.

[0100] 5) Construct an interpolation curve using the interpolation matrix and control points, ultimately obtaining a curve passing through n type value points Q. i A cubic B-spline curve (i = 0, 1, ..., n-1).

[0101] 6) Optimize and adjust the fitted curve as needed, by adjusting node positions, weight allocation, etc., to improve the curve fitting results.

[0102] c. To further reduce the fluctuation of the centerline, the Savitzky-Golay filtering algorithm is used to smooth and denoise the centerline. During the filtering process, the filter window width and the order of the polynomial fitting determine the filtering effect. The specific steps are as follows:

[0103] 1) Let the sliding window size be 2M+1, and the original data points in the window be x(n), where n takes the value [-M,...,0,...,M]. Construct a polynomial of order N for fitting:

[0104]

[0105] In the formula, k is the order of the polynomial, and d k are the coefficients of the polynomial.

[0106] 2) Least squares fitting residuals

[0107]

[0108] 3) To minimize E, let the derivative of E with respect to each coefficient be 0, i.e.

[0109]

[0110] The polynomial coefficients d can be obtained. k .

[0111] 4) Each time the sliding window is moved by 2M+1, the value of this polynomial at the center point of the window is the filtered result. When the x-coordinate is 0, the filtered value is f(0)=d0.

[0112] An application embodiment of the present invention provides a computer device, which includes a memory and a processor. The memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the EAST-based line laser stripe center extraction method.

[0113] An application embodiment of the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, causes the processor to perform the steps of a line laser stripe center extraction method based on EAST.

[0114] An application embodiment of the present invention provides an information data processing terminal, which is used to implement a line laser stripe center extraction system based on EAST.

[0115] The proposed embodiments have achieved some positive results during the research and development or use process, and have indeed great advantages compared with the existing technology. The following content describes them in conjunction with the data, charts and other information from the experimental process.

[0116] This invention constructs a deep learning-based laser stripe segmentation algorithm, effectively removing interference from environmental noise and uneven light reflection that affects light stripe extraction, achieving accurate separation of the light stripe region and the background. Furthermore, based on the traditional grayscale centroid method, it introduces cubic B-spline curve fitting to reconnect discontinuous center lines, and combines this with a Savitzky-Golay filtering algorithm for smoothing and denoising the center lines, ensuring the accuracy and real-time performance of light stripe center extraction. Simultaneously, this invention exhibits superior characteristics in terms of extraction accuracy, processing speed, and noise resistance.

[0117] The data in Table 1 shows that the standard error of the algorithm presented in this paper is the smallest compared to the other two algorithms, thus the algorithm presented in this paper has the highest accuracy. The Steger algorithm has the largest standard error of light stripes, and the accuracy of the traditional gray-scale centroid method is also slightly lower than that of the algorithm presented in this paper.

[0118] Table 1. Error in laser stripe centerline extraction

[0119]

[0120] Observing the data in Table 2, we can see that the average processing time of the traditional gray-scale centroid method is approximately 0.053s, while the average processing time of the method proposed in this paper is 0.050s. This shows that the processing time of the proposed method is not significantly different from that of the traditional gray-scale centroid method. The average processing time of the Steger algorithm is approximately 0.268s, meaning the processing time of the proposed algorithm is 20 times faster than that of the Steger algorithm. However, the Steger method requires a large number of convolution operations, resulting in a high computational cost and low efficiency in extracting the center line.

[0121] Table 2 shows the running time of each algorithm.

[0122]

[0123]

[0124] Example 1:

[0125] In a factory, a line laser stripe center extraction method based on EAST was used to rapidly measure the 3D shape of a certain part. Line structured light stripes were projected using a laser projector, and then a calibrated camera was used to acquire the laser stripes on the part's surface. Median filtering was then applied for noise reduction, and the EAST network was trained using manually labeled images to separate the light stripes from the background. Next, the sub-pixel coordinates were extracted using the gray-scale centroid method, and the initial centerline was fitted using a cubic B-spline curve. Finally, Savitzky-Golay filtering was used to optimize the centerline, achieving accurate extraction. This method successfully enabled rapid 3D shape measurement of the part.

[0126] Example 2:

[0127] In an underwater resource exploration project, a line laser stripe center extraction method based on EAST was adopted to obtain 3D information of underwater topography. First, structured light stripes were projected onto the underwater topographic surface using a laser projector, and then a calibrated camera was used to capture the laser stripes. Next, median filtering was applied for noise reduction, and the EAST network was trained using manually labeled images to separate the light stripes from the background. Then, the sub-pixel coordinates were extracted using the gray-scale centroid method, and the initial centerline was fitted using a cubic B-spline curve. Finally, Savitzky-Golay filtering was used to optimize the centerline, achieving accurate extraction. This method successfully obtained 3D information of the underwater topography, providing important data for underwater resource exploration.

[0128] It should be noted that embodiments of the present invention can be implemented in hardware, software, or a combination of both. The hardware portion can be implemented using dedicated logic; the software portion can be stored in memory and executed by a suitable instruction execution system, such as a microprocessor or dedicated-design hardware. Those skilled in the art will understand that the above-described devices and methods can be implemented using computer-executable instructions and / or included in processor control code, for example, such code provided on a carrier medium such as a disk, CD, or DVD-ROM, a programmable memory such as read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. The devices and modules of the present invention can be implemented by hardware circuitry such as very large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, or programmable hardware devices such as field-programmable gate arrays, programmable logic devices, etc., or by software executed by various types of processors, or by a combination of the above-described hardware circuitry and software, such as firmware.

[0129] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications, equivalent substitutions, and improvements made by those skilled in the art within the scope of the technology disclosed in the present invention, and within the spirit and principles of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for extracting the center of a line laser stripe based on EAST, characterized in that, include: First, the laser stripe image is acquired using a calibrated camera, and then the image is denoised using median filtering to remove isolated noise points. The EAST network was trained using manually labeled images to identify light stripe targets and separate the light stripes from the background. Then, the sub-pixel coordinates of the light stripes were extracted using the traditional gray-scale centroid method. The initial center line was fitted with a cubic B-spline curve to effectively solve the problem of discontinuity in the center line of some light stripes. Finally, Savitzky-Golay filtering was used to optimize the center line to achieve accurate extraction of the light stripe center line. The overall framework for image segmentation and detection includes: employing a scene text detection algorithm, namely the EAST detection algorithm; designing a deep neural network; manually annotating the effective region of the stripes and the original laser image; using these annotations as inputs to train the EAST deep learning model; using the trained model to detect the light stripe image under test, directly predicting the light stripe region in the overall image; then using non-maximum suppression to filter out redundant bounding boxes; and employing backpropagation algorithm to quickly converge the network to obtain the trained model, thus obtaining the final result. The final output is the coordinates of the four points of the light stripe bounding box.

2. The method for extracting the center of a line laser stripe based on EAST as described in claim 1, characterized in that, Central extraction includes: Step 1: Extract the initial center line using the grayscale centroid method; Step 2: Reconnect the broken lines using cubic B-spline curve fitting; Step 3: The Savitzky-Golay filtering algorithm is used to smooth and denoise the center line.

3. The method for extracting the center of a line laser stripe based on EAST as described in claim 1, characterized in that, In step one, the gray-scale centroid method sorts the gray values ​​of the light stripe within the region of interest, extracts the centroid of the gray values ​​of each row / column of each pixel in the cross-section of the light stripe, and uses the gray values ​​as weights to represent the centerline of the light stripe. The calculation formula is as follows: In the formula (x i ,y i F represents the coordinates of a pixel in a row or column of the light stripe. ( x i ,y i) The grayscale value corresponds to the coordinate. M F(x) represents the number of pixels on the cross-section of the light stripe. i ,y i ) represents the coordinates of the center point of the light stripe obtained from this cross section.

4. The method for extracting the center of a line laser stripe based on EAST as described in claim 1, characterized in that, In step two, the formula for the B-spline function is as follows: Where C i (i = 0, 1, ..., n) are the control vertices, N i,k (t)(i=1,2...,n) is a k-order B-spline basis function; Where the basis function N i,k (t) is: The equation of the cubic B-spline curve is: G(t)=G0*N 0,3 (t)+G1*N 1,3 (t)+G2*N 2,3 (t)+G3*N 3,3 (t); 1) First, n type value points Q are given. i (i = 0, 1, ... n-1), that is, there are n-1 cubic B-spline curves; 2) Based on the given shape points, find the vertices of the corresponding B-spline curve, shape points {Q}. i } and control point {C i The relationship between} is (C i-1 +4C i +C i+1 ) / 6=Q i (i = 0, 1, ... n-1), this system of equations has n equations and n+2 unknowns, so the boundary conditions at the endpoints need to be added; 3) Determine the boundary conditions for the cubic spline curve. For non-closed curves, take... By combining the above n equations, the control point C can be determined. i The position coordinates of (i = 0, 1, ..., n-1); 4) Construct the interpolation matrix: Construct the interpolation matrix of the cubic B-spline curve. The interpolation matrix is ​​a 4x4 matrix, which is calculated based on the basis functions of the B-spline curve. 5) Construct an interpolation curve using an interpolation matrix and control points, ultimately obtaining a curve passing through n type value points Q. i Cubic B-spline curves for (i = 0, 1, ..., n-1); 6) Optimize and adjust the fitted curve as needed, by adjusting the node positions and weight allocation, to improve the curve fitting results.

5. The method for extracting the center of a line laser stripe based on EAST as described in claim 1, characterized in that, Step three includes: In the filtering process, the filter window width and the order of the polynomial fitting determine the filtering effect. The specific steps are as follows: 1) Let the sliding window size be 2M+1, and the original data points in the window be x(n), where n takes the value [-M,...,0,...,M]. Construct a polynomial of order N for fitting: In the formula, k is the order of the polynomial, and d k The coefficients of the polynomial; 2) Least squares fitting residuals 3) To minimize E, let the derivative of E with respect to each coefficient be 0, i.e. The polynomial coefficients d can be obtained. k ; 4) Each time the sliding window is moved by 2M+1, the value of this polynomial at the center point of the window is the filtered result. When the x-coordinate is 0, the filtered value is f(0)=d0.

6. An EAST-based line laser fringe center extraction system applying the EAST-based line laser fringe center extraction method as described in claims 1-5, the EAST-based line laser fringe center extraction system comprising: The image acquisition module is used to acquire images of laser stripes. The image preprocessing module is used for median filtering and noise reduction, and for ROI region extraction using the EAST deep learning model. The centerline extraction module is used to extract the initial centerline using the grayscale centroid method; The centerline fitting module is used to fit the initial centerline using cubic B-splines; The centerline optimization module is used to optimize the centerline using Savitzky-Golay filtering.

7. A computer device comprising a memory and a processor, the memory storing a computer program, which, when executed by the processor, causes the processor to perform the steps of the EAST-based line laser stripe center extraction method as described in claims 1 to 5.

8. A computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the steps of the EAST-based line laser stripe center extraction method as described in claims 1 to 5.

9. An information data processing terminal, the information data processing terminal being used to implement the EAST-based line laser stripe center extraction system as described in claim 6.

Images