A fast 3D surface topography stitching method based on improved ORB

Abstract

The present application discloses a method for rapid splicing of three-dimensional surface morphology based on an improved ORB, comprising: determining the overlapping area of ​​the micro-nano surface structure to be spliced ​​based on a block matching method; extracting an initial feature point set based on the FAST algorithm, calculating the Harris corner response value of the feature points after non-maximum suppression processing, and screening according to the response value; describing multiple feature points to be matched with the largest response value to obtain feature descriptors; obtaining coarse matching point pairs based on the feature descriptors of the feature points to be matched, performing fine matching on the coarse matching point pairs using a decoupled RANSAC algorithm, obtaining a three-dimensional space rigid body transformation model, and performing splicing and fusion of micro-nano surface structures through transformation interpolation and overlapping area data fusion. The present application also discloses a device for rapid splicing of three-dimensional surface morphology based on an improved ORB, corresponding equipment, and storage medium. The present application can significantly improve splicing efficiency and accuracy, and effectively solve the problem of scale differences in micro-nano surface structure data in different directions.

Description

Technical Field

[0001] The present application relates to the technical field of three-dimensional shape measurement, and more specifically, to a three-dimensional surface shape rapid splicing method based on an improved ORB. Background Art

[0002] In modern industry and scientific research, the precise measurement and characterization of an object's surface topography is of great significance. Surface topography includes surface roughness, waviness, and structure, characteristics that significantly impact the performance and functionality of objects in fields such as engineering, biomedicine, and materials science. With the advancement of technology, non-contact three-dimensional topography measurement technology has gradually become mainstream, offering advantages such as fast measurement speed, high accuracy, and no damage to the surface of the object being measured. However, due to the limited field of view of the measuring equipment, a single measurement often cannot cover the surface topography of a large area. Therefore, it is necessary to use stitching technology to combine the data from multiple measurements into a complete three-dimensional topography image.

[0003] At present, three-dimensional surface topography stitching technology is mainly divided into stitching methods based on auxiliary equipment and free stitching methods. Although the stitching method with the help of auxiliary equipment has a simple algorithm, the system flexibility is poor and the cost is high. Although the free stitching method is flexible to use, it is prone to cumulative errors. In addition, when dealing with micro-nano structures, the existing stitching methods often have difficulty in balancing stitching efficiency and accuracy due to the complexity of their height information and the large amount of data. For example, traditional image stitching methods are only applicable to two-dimensional images, and it is difficult to effectively deal with the stitching of micro-nano surface structure height maps, which easily causes height information distortion. Therefore, it is particularly urgent to develop an efficient, high-precision stitching technology suitable for micro-nano surface structures. Summary of the Invention

[0004] In response to at least one defect or improvement need in the prior art, the present application provides a method for rapid splicing of three-dimensional surface morphologies based on an improved ORB, which can solve at least one of the problems existing in the above-mentioned background technology.

[0005] To achieve the above objectives, according to a first aspect of the present application, a method for rapid splicing of three-dimensional surface morphologies based on an improved ORB is provided, the method comprising:

[0006] Based on the block matching method, a lateral sliding search is performed on the micro-nano surface structure to be spliced ​​to determine the overlapping area;

[0007] In the overlapping area, an initial feature point set is extracted based on the FAST algorithm, non-maximum suppression is performed on the initial feature points, redundant feature points are removed, Harris corner point response values ​​of the feature points after non-maximum suppression are calculated, and screening is performed based on the response values ​​to obtain multiple feature points with the largest response values ​​to be matched;

[0008] Describing the plurality of feature points to be matched with the largest response values ​​to obtain feature descriptors;

[0009] Based on the feature descriptors of the feature points to be matched, the feature points to be matched are filtered using a K-nearest neighbor search algorithm to obtain coarse matching point pairs, and the coarse matching point pairs are finely matched using a decoupled RANSAC algorithm to obtain a three-dimensional rigid body transformation model;

[0010] Based on the three-dimensional rigid body transformation model, micro-nano surface structure splicing and fusion are performed through transformation interpolation and overlapping area data fusion.

[0011] Furthermore, the above-mentioned three-dimensional surface morphology fast stitching method based on the improved ORB, extracting an initial feature point set based on the FAST algorithm, performing non-maximum suppression processing on the initial feature points, and removing redundant feature points, specifically includes:

[0012] For each coordinate point in the overlapping area , whose height is , take this point as the center point, select the first number on the circle of the first radius m coordinate points ;

[0013] Preset threshold t If there are a second number of consecutive coordinate points whose height difference with the center point is greater than the first threshold t ,Right now

[0014]

[0015] Then the coordinate point As candidate feature points;

[0016] For candidate feature points , calculate its response value

[0017]

[0018] The candidate feature points Response value Compared with the response values ​​of all other candidate feature points within the range, if Response value is the local maximum, then keep is the initial feature point, otherwise it will be suppressed.

[0019] Furthermore, the above-mentioned three-dimensional surface morphology fast stitching method based on the improved ORB, wherein the Harris corner response value is calculated, and screening is performed according to the response value to obtain multiple feature points to be matched with the largest response value, specifically includes:

[0020] Height map of the overlapping area , calculated by the Prewitt operator exist 、 Gradient in direction

[0021]

[0022] in, represents convolution;

[0023] For each initial coordinate point, calculate the local autocorrelation matrix

[0024]

[0025] in, is a local window of size ;

[0026] Calculate Harris corner response values

[0027]

[0028] in, is a matrix The determinant of is a matrix traces, is an empirical constant;

[0029] Sort the calculated corner point response values ​​and retain multiple feature points to be matched with the largest response values.

[0030] Furthermore, the above-mentioned three-dimensional surface morphology rapid splicing method based on the improved ORB, describing the multiple feature points to be matched with the largest response values ​​to obtain feature descriptors, specifically includes:

[0031] For the current feature point to be matched, determine a circle with a size of The local neighborhood of

[0032] Generate in the local neighborhood For random coordinate point pairs ;

[0033] For each pair of random coordinate points , compare their values ​​in the micro-nano surface structure height map and ,have

[0034]

[0035] All The comparison results of the coordinate point pairs are connected in sequence to form a length of The binary vector is the feature descriptor of the feature point

[0036] .

[0037] Furthermore, the above-mentioned three-dimensional surface morphology fast stitching method based on the improved ORB, based on the feature descriptors of the feature points to be matched, filters the feature points to be matched by the K nearest neighbor search algorithm to obtain rough matching point pairs, specifically including:

[0038] For the set of feature points to be matched, the normalized Hamming distance is used to measure the similarity of the features.

[0039]

[0040] in, and It is a 256-bit feature descriptor;

[0041] For the heightmap Feature points , in the height map Find the two nearest neighbor points of the feature points 、 , Feature points and in The two nearest neighbors in 、 ;

[0042] Cross-validation and Is it a rough matching point pair, including verification and The nearest neighbors of are each other, the normalized Hamming distance of the nearest neighbor is less than the threshold, and the distance ratio between the nearest neighbor and the second nearest neighbor is less than the threshold, that is,

[0043]

[0044] in, 、 、 They are 、 、 feature descriptors.

[0045] Furthermore, the above-mentioned three-dimensional surface morphology fast stitching method based on the improved ORB, wherein the decoupled RANSAC algorithm is used to perform fine matching on the rough matching point pairs to obtain a three-dimensional rigid body transformation model, specifically includes:

[0046] from Randomly select from the rough matching points , calculate the two-dimensional plane rigid body transformation model, use the model to map all point pairs, calculate the horizontal Euclidean distance as the reprojection error, if the reprojection error is less than the first preset threshold, then the point pair is considered as an inlier. Through multiple iterative calculations, the two-dimensional optimal inlier point set that satisfies horizontal alignment is obtained. The number of inliers is ;

[0047] from Randomly select from two-dimensional optimal interior points Calculate the 3D planar rigid body transformation model, use the model to map all point pairs, calculate the Z-axis distance of corresponding point pairs as the reprojection error, and if the reprojection error is less than a second preset threshold, then consider the point pair as an inlier. Through multiple iterative calculations, obtain the 3D optimal inlier point set that satisfies spatial alignment;

[0048] Based on the three-dimensional optimal interior point set, the three-dimensional rigid body transformation model is fitted by the least squares method.

[0049] .

[0050] Furthermore, the above-mentioned three-dimensional surface morphology fast stitching method based on the improved ORB, based on the three-dimensional space rigid body transformation model, performs micro-nano surface structure stitching and fusion through transformation interpolation and overlapping area data fusion, specifically including:

[0051] Based on the bilinear interpolation principle, the Z-axis height of the interpolation point O is calculated as follows:

[0052]

[0053] in, 、 、 、 They are the height values ​​of the four mapping points around the point to be interpolated;

[0054] After completing the interpolation and registration of the data, the overlapping areas are fused through weighted fusion. Weighted fusion assigns different weights to the two sets of data based on the distance between the feature points and the edge of the overlapping area. The weight of the overlapping area is:

[0055]

[0056] in, is the total width of the overlapping area, For data points Distance from the left border;

[0057] Micro-nano surface structure data and Perform splicing and fusion, the splicing and fusion result is ,have

[0058] .

[0059] According to a second aspect of the present application, a three-dimensional surface topography rapid splicing device based on an improved ORB is also provided, comprising:

[0060] A block matching module is used to perform a lateral sliding search on the micro-nano surface structure to be spliced ​​based on a block matching method to determine the overlapping area;

[0061] A feature extraction module is used to extract an initial feature point set in the overlapping area based on the FAST algorithm, perform non-maximum suppression on the initial feature points, remove redundant feature points, calculate Harris corner point response values ​​of the feature points after the non-maximum suppression process, and screen them according to the response values ​​to obtain multiple feature points with the largest response values ​​to be matched;

[0062] A descriptor acquisition module, configured to describe the plurality of feature points to be matched with the largest response values ​​and obtain feature descriptors;

[0063] A matching module is used to filter the feature points to be matched using a K-nearest neighbor search algorithm based on the feature descriptors of the feature points to obtain coarse matching point pairs, and to perform fine matching on the coarse matching point pairs using a decoupled RANSAC algorithm to obtain a three-dimensional rigid body transformation model;

[0064] The splicing and fusion module is used to perform splicing and fusion of micro-nano surface structures through transformation interpolation and overlapping area data fusion based on the three-dimensional space rigid body transformation model.

[0065] According to the third aspect of the present application, a three-dimensional surface morphology rapid splicing device based on an improved ORB is also provided, which includes at least one processing unit and at least one storage unit, wherein the storage unit stores a computer program, and when the computer program is executed by the processing unit, the processing unit performs the steps of any one of the above methods.

[0066] According to the fourth aspect of the present application, a storage medium is also provided, which stores a computer program that can be executed by a three-dimensional surface morphology rapid splicing device based on an improved ORB. When the computer program runs on the three-dimensional surface morphology rapid splicing device based on an improved ORB, the three-dimensional surface morphology rapid splicing device based on an improved ORB executes the steps of any one of the methods described above.

[0067] In general, the above technical solutions conceived by this application can achieve the following beneficial effects compared with the existing technology:

[0068] The embodiment of the present application provides a three-dimensional surface morphology fast splicing method based on an improved ORB, which can effectively reduce the number of feature points and improve the quality of feature points by combining the FAST algorithm and the Harris corner detection algorithm for feature extraction, thereby significantly improving the splicing efficiency and accuracy. At the same time, by performing fine matching of coarse matching point pairs through the decoupled RANSAC algorithm, the scale difference problem of micro-nano surface structure data in the horizontal and vertical directions can be solved, further improving the accuracy and reliability of matching. In addition, based on the transformation interpolation and overlapping area data fusion of the three-dimensional space rigid body transformation model, efficient splicing and fusion of micro-nano surface structures can be achieved, ensuring smooth transition and high-precision alignment of the splicing results. By optimizing the feature extraction and fine matching process, the splicing efficiency and accuracy can be significantly improved, the scale difference problem of micro-nano surface structure data in different directions can be effectively solved, and efficient and high-precision surface morphology splicing can be achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0070] Figure 1 A schematic flow chart of a method for rapid splicing of three-dimensional surface morphologies based on an improved ORB provided in an embodiment of the present application. DETAILED DESCRIPTION

[0071] In order to make the purpose, technical solutions and advantages of this application more clearly understood, this application is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are merely for explaining this application and are not intended to limit this application. In addition, the technical features involved in the various embodiments of this application described below may be combined with each other as long as they do not conflict with each other.

[0072] The terms "first," "second," "third," and the like in the specification and claims of this application and the accompanying drawings are used to distinguish between different objects, not to describe a particular order. Furthermore, the terms "including," "having," and any variations thereof, are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or apparatus comprising a series of steps or elements is not limited to the listed steps or elements, but may optionally include steps or elements not listed, or may optionally include other steps or elements inherent to the process, method, product, or apparatus.

[0073] Figure 1A schematic diagram of a process for rapid splicing of three-dimensional surface morphologies based on an improved ORB is provided in an embodiment of the present application, as shown in FIG. Figure 1 As shown, the embodiment of the present application provides a method for rapid splicing of three-dimensional surface morphologies based on an improved ORB, comprising:

[0074] S1 performs a lateral sliding search on the micro-nano surface structure to be spliced ​​based on the block matching method to determine the overlapping area;

[0075] S2 extracts an initial feature point set within the overlapping area based on the FAST algorithm, performs non-maximum suppression on the initial feature points, removes redundant feature points, calculates Harris corner point response values ​​of the feature points after the non-maximum suppression process, and screens them based on the response values ​​to obtain multiple feature points with the largest response values ​​to be matched;

[0076] S3 describes the multiple feature points to be matched with the largest response values ​​to obtain feature descriptors;

[0077] S4: Based on the feature descriptors of the feature points to be matched, filtering the feature points to be matched by a K-nearest neighbor search algorithm to obtain coarse matching point pairs, and performing fine matching on the coarse matching point pairs by a decoupled RANSAC algorithm to obtain a three-dimensional rigid body transformation model;

[0078] S5 is based on the three-dimensional rigid body transformation model and performs micro-nano surface structure splicing and fusion through transformation interpolation and overlapping area data fusion.

[0079] Specifically, the micro-nano surface structure stitching algorithm proposed in this application introduces the ORB algorithm. In order to ensure a high stitching efficiency, the feature extraction and feature matching algorithms are improved according to the characteristics of white light interferometry data. In the feature extraction stage, the FAST algorithm in ORB is used for feature extraction. The number of feature points is too large and the algorithm efficiency is reduced. This application proposes a feature extraction algorithm that combines FAST and Harris. The feature points extracted by FAST are screened according to the Harris response value of the Harris response value, the number of feature points is controlled, and the algorithm efficiency is improved. In the precise matching process of feature matching, the standard RANSAC algorithm is inaccurate and fails due to the inconsistent horizontal and vertical scales of the micro-nano surface structure data obtained by white light interferometry. This application proposes a decoupled RANSAC algorithm to decouple the three-dimensional error and divide it into the Euclidean distance error in the horizontal direction and the height error in the vertical direction, which correspond to the two stages of two-dimensional horizontal registration and high-precision vertical height registration, respectively, to eliminate the interference caused by the difference in data horizontal and vertical accuracy.

[0080] The improved ORB-based stitching of micro-nano surface structures mainly includes the following steps: determining overlapping areas, feature extraction and description, feature matching, and stitching and fusion. First, for the micro-nano surface structures to be stitched, the overlapping areas are preliminarily determined using the block matching method. Next, FAST and Harris are combined to extract features from the overlapping areas and perform BRIEF description on the detected feature points. Then, preliminary matching point pairs are obtained through coarse matching, and the point pairs are screened through decoupled RANSAC fine matching. The final matching point pairs are used to calculate the transformation model. Finally, the transformation model is used to achieve the stitching and fusion of the micro-nano surface structures.

[0081] In one embodiment, a block matching method may be used to determine the overlapping areas, e.g. and are two adjacent micro-nano surface structure height maps on the left and right, and the number of sampling points is ,exist Select a size Block , the block size needs to be chosen reasonably to ensure exist In the overlapping area of Slide horizontally to search and Blocks in The mean square error (MSE) of

[0082]

[0083] in, and Block and The point in .

[0084] according to and The position of the overlap area can be quickly determined :

[0085]

[0086] in, For blocks and The distance to the right border, For blocks and The distance from the left border, The width of the block.

[0087] Since only horizontal sliding search is involved, the time complexity of the block matching method to determine the overlapping area is:

[0088]

[0089] In the case of full search, the time complexity is:

[0090]

[0091] The computational complexity of horizontal sliding search is only , which can significantly reduce the amount of calculation, especially when the vertical search range is large, the difference is more significant. For micro-nano surface structures with mainly lateral displacement, the lateral search block matching method is better than the full search block matching method.

[0092] The embodiment of the present application provides a three-dimensional surface morphology fast splicing method based on an improved ORB, which can effectively reduce the number of feature points and improve the quality of feature points by combining the FAST algorithm and the Harris corner detection algorithm for feature extraction, thereby significantly improving the splicing efficiency and accuracy. At the same time, by performing fine matching of coarse matching point pairs through the decoupled RANSAC algorithm, the scale difference problem of micro-nano surface structure data in the horizontal and vertical directions can be solved, further improving the accuracy and reliability of matching. In addition, based on the transformation interpolation and overlapping area data fusion of the three-dimensional space rigid body transformation model, efficient splicing and fusion of micro-nano surface structures can be achieved, ensuring smooth transition and high-precision alignment of the splicing results. By optimizing the feature extraction and fine matching process, the splicing efficiency and accuracy can be significantly improved, the scale difference problem of micro-nano surface structure data in different directions can be effectively solved, and efficient and high-precision surface morphology splicing can be achieved.

[0093] Optionally, the improved ORB-based three-dimensional surface topography fast stitching method provided in an embodiment of the present application extracts an initial feature point set based on the FAST algorithm, performs non-maximum suppression processing on the initial feature points, and removes redundant feature points, specifically including:

[0094] For each coordinate point in the overlapping area , whose height is , take this point as the center point, select the first number on the circle of the first radius m coordinate points ;

[0095] Preset threshold t If there are a second number of consecutive coordinate points whose height difference with the center point is greater than the first threshold t ,Right now

[0096]

[0097] Then the coordinate point As candidate feature points;

[0098] For candidate feature points , calculate its response value

[0099]

[0100] The candidate feature points Response value Compared with the response values ​​of all other candidate feature points within the range, if Response value is the local maximum, then keep is the initial feature point, otherwise it will be suppressed.

[0101] Specifically, key features such as corners, edges, and textures are extracted from micro-nano surface structure data. The significance of these features directly affects the quality of subsequent matching and stitching. Compared to dense texture and edge features, corners are fewer in number but contain richer information, which can significantly reduce the amount of computation and improve algorithm efficiency. Corner points generally correspond to local extreme values ​​and represent significant changes in surface structure data. When a point differs significantly from multiple surrounding points in multiple directions, it indicates that the point is at the intersection of multiple edges or texture changes and can be considered a corner point.

[0102] The FAST (Features from Accelerated Segment Test) algorithm is a fast corner detection method based on numerical comparison of coordinate points. It determines whether a point is a corner by comparing the numerical differences between the center point and the surrounding coordinate points. The specific process of the FAST algorithm is as follows:

[0103] For the coordinate points in the overlapping area of ​​the micro-nano surface structure height map , whose height is . As the center, select 16 coordinate points on a circle with a radius of 3 , the corresponding height value is . Set a threshold , if there is a continuous coordinate points, satisfying

[0104]

[0105] or

[0106]

[0107] Then the coordinate point is considered as a candidate corner point.

[0108] The main parameters of the FAST algorithm are threshold , the number of continuous coordinate points in corner point judgment , will have a direct impact on the corner point determination. Selecting a larger threshold can suppress the noise effect to a certain extent and improve the stability of the corner points, but it will cause some real corner points to be missed. It can be determined based on the root mean square of the micro-nano surface structure height map:

[0109]

[0110]

[0111] in, is a constant with an empirical range of 0.2 to 0.4. RMS is the root mean square of the overlapped area. 、 are the width and height of the overlapping area, for The height value at is the mean height of the overlapping area.

[0112] Number of consecutive coordinate points in corner point judgment Determines the strictness of FAST corner point determination, the larger It can improve the reliability of corner detection and reduce the number of overall corner points, but it will make the algorithm more sensitive to noise, and it is easy for noise to suppress the detection of real corner points. It will also increase the amount of calculation and reduce the efficiency of the algorithm. Taking 9 is a balance between speed and accuracy.

[0113] The FAST algorithm achieves efficient corner detection by quickly comparing the numerical values ​​of coordinate points. However, corner points themselves are local area features rather than isolated points, such as the corners of objects and edge intersections. In actual detection, coordinate points that meet the corner point determination conditions are often clustered together, and multiple adjacent corner points will be detected in the same area, resulting in redundant corner point problems. Redundant corner points increase the computational complexity of subsequent matching and also lead to inaccurate feature positioning, affecting matching accuracy. To solve the problem of redundant corner points, non-maximum suppression (NMS) is performed after the FAST algorithm. By retaining the corner point with the largest local response value and suppressing other candidate points in its neighborhood, redundancy is removed, the positioning accuracy of the corner points is improved, and the efficiency and accuracy of subsequent matching are improved. The process of non-maximum suppression is as follows:

[0114] For candidate feature points , and its response value can be expressed as:

[0115]

[0116] It reflects the difference between the candidate feature point and the surrounding coordinate points. The greater the difference, the greater the response value, and the more likely it is a true corner point. In the neighborhood of Compare with the response values ​​of all candidate points in the range, if If the response value is the local maximum, it is retained as the FAST feature point, otherwise it is suppressed.

[0117] The size of the neighborhood for non-maximum suppression affects the number of feature points. Selecting a larger neighborhood can effectively reduce the number of feature points and suppress redundant corner points. However, a larger neighborhood can lead to incorrect suppression of valid corner points and miss detection of true feature points. The neighborhood size for non-maximum suppression should not be too large; typically, it should be 3×3 or 5×5.

[0118] The FAST algorithm needs to process every point in the overlapping area, and each point is compared at most 16 times, so the time complexity is:

[0119]

[0120] in, 、 are the width and height of the overlapping area respectively.

[0121] Optionally, the improved ORB-based three-dimensional surface topography fast stitching method provided in an embodiment of the present application, wherein the Harris corner response values ​​are calculated, and screening is performed based on the response values ​​to obtain multiple feature points to be matched with the largest response values, specifically including:

[0122] Height map of the overlapping area , calculated by the Prewitt operator exist 、 Gradient in direction

[0123]

[0124] in, represents convolution;

[0125] For each initial coordinate point, calculate the local autocorrelation matrix

[0126]

[0127] in, is a local window of size ;

[0128] Calculate Harris corner response values

[0129]

[0130] in, is a matrix The determinant of is a matrix traces, is an empirical constant;

[0131] Sort the calculated corner point response values ​​and retain multiple feature points to be matched with the largest response values.

[0132] Specifically, the Harris algorithm is a corner detection algorithm based on the height change of the coordinate point neighborhood. It monitors the height change through gradient information. If a point causes a significant change in height value in all directions, then this point is very likely to be a corner point.

[0133] The height map of the overlapping area of ​​the micro-nano surface structure to be spliced ​​is , calculated by the Prewitt operator exist 、 The gradient in the direction, for each coordinate point in the overlapping area, calculate the local autocorrelation matrix ,get Then, calculate the Harris corner response value, where is an empirical constant (usually between 0.04 and 0.06).

[0134] The larger the Harris corner response value, the 、 The more significant the gradient change in a direction, the greater the possibility of it being a corner point. Harris analyzes the height change in two directions by calculating the autocorrelation matrix of the local window, suppressing the influence of noise. Compared with the FAST algorithm which is easily affected by noise, the Harris algorithm has higher corner point extraction accuracy. The Harris algorithm needs to calculate the autocorrelation matrix of all points in the overlapping area. 、 The gradient in the direction corresponds to , the window summation needs to be calculated when the autocorrelation matrix is ​​obtained, so the total time complexity is:

[0135]

[0136] in, 、 are the width and height of the overlapping area, is the Harris local window size.

[0137] The FAST algorithm and non-maximum suppression are used to quickly obtain a set of candidate feature points. The Harris corner response values ​​of the candidate feature points are calculated, sorted according to the response values, and the feature points with the largest response values ​​are retained. This method balances computational efficiency and feature point quality. It uses the faster FAST algorithm for global calculations, avoiding the huge computational burden of directly applying the Harris algorithm to process micro-nano surface structure data. It also improves the reliability of feature points through secondary screening using the Harris algorithm.

[0138] The time complexity of the feature extraction algorithm combining FAST and Harris is:

[0139]

[0140] in, 、 are the width and height of the overlapping area, is a constant ( ), and are the number of feature points before and after non-maximum suppression, is the Harris local window size.

[0141] The time complexity of the feature extraction algorithm combining FAST and Harris is slightly higher than that of the FAST algorithm, but it can reduce the number of feature points while ensuring the quality of feature points, which can greatly improve the efficiency of subsequent matching and improve the overall efficiency of the algorithm.

[0142] Optionally, the improved ORB-based three-dimensional surface topography rapid stitching method provided in an embodiment of the present application, wherein the description of the plurality of feature points to be matched with the largest response values ​​to obtain feature descriptors specifically includes:

[0143] For the current feature point to be matched, determine a circle with a size of The local neighborhood of

[0144] Generate in the local neighborhood For random coordinate point pairs ;

[0145] For each pair of random coordinate points , compare their values ​​in the micro-nano surface structure height map and ,have

[0146]

[0147] All The comparison results of the coordinate point pairs are connected in sequence to form a The binary vector is the feature descriptor of the feature point

[0148] .

[0149] Specifically, feature extraction aims to identify significant feature points in micro-nano surface structures. However, these feature points only contain positional information and cannot meet the requirements of feature matching. Therefore, a unique descriptor, or feature descriptor, must be generated for each detected feature point. Feature descriptors typically exist in the form of numerical vectors, representing the structural information of the local region surrounding the feature point, thus making the feature point unique. Common feature description methods include gradient-based, binary-based, and histogram-based methods.

[0150] The BRIEF descriptor generates a binary vector by comparing a series of random coordinate point pairs to describe the local information around the feature point. It does not require complex calculations and is convenient and efficient. The specific process of obtaining the BRIEF descriptor of a feature point is as follows:

[0151] For the current feature point, determine a size around it The size of the neighborhood needs to be weighed according to the specific application scenario, and it is necessary to ensure that sufficient local information can be captured while avoiding introducing too much noise.

[0152] In the local neighborhood, generate For random pairs of coordinate points: , the positions of these coordinate point pairs relative to the central feature point are fixed.

[0153] For each pair of random coordinate points , compare their values ​​in the micro-nano surface structure height map and .if Greater than , then the comparison result of this pair of coordinate points is Otherwise, the comparison result is recorded as 0:

[0154]

[0155] Construction of binary descriptor: All The comparison results of the coordinate point pairs are connected in sequence to form a length of The binary vector is the BRIEF descriptor of the feature point:

[0156]

[0157] The longer the BRIEF descriptor is, the higher the discrimination is, but the matching calculation is more complex. The shorter the descriptor is, the faster it is, but the matching accuracy may be reduced. In order to balance accuracy and efficiency, here Take 8, that is, the length of the BRIEF feature descriptor is 256, and correspondingly, the size of the local neighborhood can be set to 31×31.

[0158] Optionally, the improved ORB-based three-dimensional surface topography fast stitching method provided in an embodiment of the present application, wherein the feature descriptors of the feature points to be matched are used to filter the feature points to be matched using a K-nearest neighbor search algorithm to obtain coarse matching point pairs, specifically including:

[0159] For the set of feature points to be matched, the normalized Hamming distance is used to measure the similarity of the features.

[0160]

[0161] in, and It is a 256-bit feature descriptor;

[0162] For the heightmap Feature points , in the height map Find the two nearest neighbor points of the feature points 、 , Feature points and in The two nearest neighbors in 、 ;

[0163] Cross-validation and Is it a rough matching point pair, including verification and The nearest neighbors of are each other, the normalized Hamming distance of the nearest neighbor is less than the threshold, and the distance ratio between the nearest neighbor and the second nearest neighbor is less than the threshold, that is,

[0164]

[0165] in, 、 、 They are 、 、 feature descriptors.

[0166] Specifically, after feature extraction, feature matching is used to find matching point pairs and establish correspondences between the regions to be joined. Feature matching is typically divided into two stages: coarse matching and fine matching. Coarse matching quickly determines the initial feature matching pairs, providing a foundation for subsequent fine matching. The fine matching stage utilizes geometric constraints to optimize the coarse matching results, eliminate mismatched points (outliers), and improve matching accuracy.

[0167] BRIEF feature descriptors represent the local information of feature points. The smaller the difference between the descriptors, the higher the matching degree of the two feature points. For the set of feature points to be matched, the normalized Hamming distance is used to measure the similarity of the features:

[0168]

[0169] in, 、 Both are 256-bit BRIEF feature descriptors. The smaller the value of and The higher the matching degree.

[0170] The coarse matching uses K nearest neighbor search and cross validation to filter out useless points quickly and efficiently. Feature points , in the height map Find the two nearest neighbor points of the feature points 、 , Feature points and in The two nearest neighbors in 、 . and For rough matching point pairs, cross validation is required, i.e. and The nearest neighbors of are each other, and at the same time, the normalized Hamming distance of the nearest neighbor must be less than the threshold, and the distance ratio between the nearest neighbor and the next nearest neighbor must be less than the threshold, that is:

[0171]

[0172] in, 、 、 They are 、 、 BRIEF descriptor.

[0173] The rough matching based on K-nearest neighbor search can be divided into building a KD-Tree, bidirectional K-nearest neighbor search and cross-validation, and the time complexity is:

[0174]

[0175] Among them, N is the number of feature points. When it is 300, the time complexity of coarse matching is very small.

[0176] Optionally, the improved ORB-based three-dimensional surface morphology fast stitching method provided in an embodiment of the present application, wherein the decoupled RANSAC algorithm is used to perform fine matching on the rough matching point pairs to obtain a three-dimensional rigid body transformation model, specifically comprising:

[0177] from Randomly select from the rough matching points , calculate the two-dimensional plane rigid body transformation model, use the model to map all point pairs, calculate the horizontal Euclidean distance as the reprojection error, if the reprojection error is less than the first preset threshold, then the point pair is considered as an inlier. Through multiple iterative calculations, the two-dimensional optimal inlier point set that satisfies horizontal alignment is obtained. The number of inliers is ;

[0178] from Randomly select from two-dimensional optimal interior points Calculate the 3D planar rigid body transformation model, use the model to map all point pairs, calculate the Z-axis distance of corresponding point pairs as the reprojection error, and if the reprojection error is less than a second preset threshold, then consider the point pair as an inlier. Through multiple iterative calculations, obtain the 3D optimal inlier point set that satisfies spatial alignment;

[0179] Based on the three-dimensional optimal interior point set, the three-dimensional rigid body transformation model is fitted by the least squares method.

[0180] .

[0181] Specifically, when feature extraction retains a large number of feature points, there may be obvious matching errors, such as large spatial distance deviations. There are also subtle matching errors, as shown in the enlarged image, where the match is close to correct but not globally consistent. To improve stitching accuracy, it is necessary to eliminate mismatched point pairs through fine matching.

[0182] The RANSAC algorithm performs well in two-dimensional image matching, but when applied to three-dimensional matching of micro-nano surface structures, its performance is affected by the characteristics of the micro-nano surface structure data. Micro-nano surface structure data can have significant accuracy differences in the vertical and horizontal directions. The RANSAC algorithm typically uses a Euclidean distance threshold to eliminate mismatched points. This single threshold setting method is difficult to simultaneously meet the accuracy differences in both directions: if a small threshold is set to meet the high-precision requirements in the vertical direction, the number of inliers in the horizontal direction will be insufficient due to the low resolution. If a large threshold is set, the matching requirements in the horizontal direction can be met, but the high-precision information in the vertical direction cannot be fully utilized, making it difficult to eliminate mismatches caused by noise.

[0183] To address the limitations of the standard RANSAC algorithm for white light interferometry data, this application proposes a decoupled RANSAC algorithm that decouples the three-dimensional error into the horizontal Euclidean distance error and the vertical height error, corresponding to the two-dimensional horizontal registration and high-precision vertical height registration stages, respectively, eliminating the interference caused by the difference in horizontal and vertical accuracy of the data. The implementation process is as follows:

[0184] from Randomly select from the rough matching points Without considering the Z-height value, a two-dimensional plane rigid body transformation model is calculated. Subsequently, the model is used to map all point pairs and the horizontal Euclidean distance is calculated as the reprojection error. If the reprojection error is less than the preset threshold, the point pair is considered as an inlier. Through multiple iterative calculations, the optimal two-dimensional inlier point set that satisfies horizontal alignment is obtained. The number of inliers is .

[0185] from Randomly select from two-dimensional optimal interior points A 3D planar rigid body transformation model is calculated. This model is then used to map all point pairs, and the Z-axis distance between corresponding point pairs is calculated as the reprojection error. If the reprojection error is less than a preset threshold, the point pair is considered an inlier. Through multiple iterative calculations, the optimal set of 3D inliers that satisfies spatial alignment is obtained.

[0186] Based on the three-dimensional optimal interior point set, the three-dimensional rigid body transformation model is fitted by the least squares method. :

[0187]

[0188] Although the decoupled RANSAC algorithm is divided into two stages, the first stage is two-dimensional registration, and the second stage is high-precision vertical height registration, which processes a small number of samples. Its time complexity is:

[0189]

[0190] in, 、 are the proportions of inliers in the two matches, 、 The minimum sample size is randomly selected for the two matches, which are 2 and 3 respectively. is the number of coarse matching point pairs.

[0191] Optionally, the improved ORB-based three-dimensional surface morphology fast stitching method provided in the embodiment of the present application, based on the three-dimensional rigid body transformation model, performs micro-nano surface structure stitching and fusion through transformation interpolation and overlapping area data fusion, specifically including:

[0192] Based on the bilinear interpolation principle, the Z-axis height of the interpolation point O is calculated as follows:

[0193]

[0194] in, 、 、 、 They are the height values ​​of the four mapping points around the point to be interpolated;

[0195] After completing the interpolation and registration of the data, the overlapping areas are fused through weighted fusion. Weighted fusion assigns different weights to the two sets of data based on the distance between the feature points and the edge of the overlapping area. The weight of the overlapping area is:

[0196]

[0197] in, is the total width of the overlapping area, For data points Distance from the left border;

[0198] Micro-nano surface structure data and Perform splicing and fusion, the splicing and fusion result is ,have

[0199] .

[0200] Specifically, after obtaining the transformation model, micro-nano surface structure stitching and fusion involves two steps: transformation interpolation and data fusion of overlapping areas. Because there is a deviation between the transformed target area data and the reference area data, interpolation is required to ensure data alignment before subsequent fusion operations. Common interpolation methods include nearest point interpolation and bilinear interpolation.

[0201] The nearest point interpolation selects the value of the known point closest to the target point as the value of the target point. It is fast, but the interpolation result may have obvious edges and discontinuities, resulting in loss of accuracy. Bilinear interpolation uses the height values ​​of the four mapped points around the interpolated point for weighted averaging. The weight is determined by the distance between the mapped coordinates and these four coordinate points. Compared with the nearest point interpolation, bilinear interpolation can produce a smoother transition, but the amount of calculation will increase. Bilinear interpolation performs linear interpolation in two directions, and then performs linear interpolation on the result to obtain an estimated value of the target point. According to the principle of bilinear interpolation, the Z-axis height of the interpolated point O is calculated as follows:

[0202]

[0203] in, 、 、 、 They are the height values ​​of the four mapping points around the point to be interpolated.

[0204] After completing the interpolation and registration of the data, the overlapping areas are fused through weighted fusion to achieve a smooth transition. Weighted fusion assigns different weights to the two sets of data based on the distance between the data points and the edge of the overlapping area. The weight of the overlapping area is:

[0205]

[0206] in, is the total width of the overlapping area, For data points Distance from the left edge.

[0207] Micro-nano surface structure data and Perform splicing and fusion, the splicing and fusion result is , For one of these points, we have:

[0208]

[0209] The RMSE (Root Mean Square Error) of the overlapped area height is used to evaluate stitching accuracy. The RMSE is an effective indicator of the deviation between the stitching result and the reference area. The RMSE of the overlapped area height is used as the evaluation criterion for stitching quality. The formula is as follows:

[0210]

[0211] in, 、 is the height of the overlapping area before and after stitching, corresponding to the reference area and the stitching fusion result respectively.

[0212] Optionally, an embodiment of the present application further provides a three-dimensional surface topography rapid splicing device based on an improved ORB, comprising:

[0213] A block matching module is used to perform a lateral sliding search on the micro-nano surface structure to be spliced ​​based on a block matching method to determine the overlapping area;

[0214] A feature extraction module is used to extract an initial feature point set in the overlapping area based on the FAST algorithm, perform non-maximum suppression on the initial feature points, remove redundant feature points, calculate Harris corner point response values ​​of the feature points after the non-maximum suppression process, and screen them according to the response values ​​to obtain multiple feature points with the largest response values ​​to be matched;

[0215] A descriptor acquisition module, configured to describe the plurality of feature points to be matched with the largest response values ​​and obtain feature descriptors;

[0216] A matching module is used to filter the feature points to be matched using a K-nearest neighbor search algorithm based on the feature descriptors of the feature points to obtain coarse matching point pairs, and to perform fine matching on the coarse matching point pairs using a decoupled RANSAC algorithm to obtain a three-dimensional rigid body transformation model;

[0217] The splicing and fusion module is used to perform splicing and fusion of micro-nano surface structures through transformation interpolation and overlapping area data fusion based on the three-dimensional space rigid body transformation model.

[0218] The present application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above method. The computer-readable storage medium may include, but is not limited to, any type of disk, including a floppy disk, an optical disk, a DVD, a CD-ROM, a microdrive, a magneto-optical disk, a ROM, a RAM, an EPROM, an EEPROM, a DRAM, a VRAM, a flash memory device, a magnetic card or an optical card, a nanosystem (including a molecular memory IC), or any type of medium or device suitable for storing instructions and / or data.

[0219] It should be noted that for the aforementioned method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that this application is not limited by the order of the actions described, because according to this application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by this application.

[0220] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0221] In the several embodiments provided in this application, it should be understood that the disclosed devices can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some service interface, and the indirect coupling or communication connection of the device or unit can be electrical or other forms.

[0222] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0223] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0224] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a memory and includes a number of instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned memory includes: U disk, read-only memory (ROM), random access memory (RAM), mobile hard disk, magnetic disk, or optical disk, etc., various media that can store program code.

[0225] Those skilled in the art will appreciate that all or part of the steps in the various methods of the above embodiments can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable memory, which may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0226] The above is only an exemplary embodiment of the present disclosure and cannot be used to limit the scope of the present disclosure. That is, any equivalent changes and modifications made according to the teachings of the present disclosure are still within the scope of the present disclosure. After considering the specification and practicing the disclosure herein, those skilled in the art will easily think of the implementation scheme of the present disclosure. This application is intended to cover any variation, use or adaptation of the present disclosure, which follows the general principles of the present disclosure and includes common knowledge or customary technical means in the art that are not recorded in the present disclosure. The description and examples are to be regarded as exemplary only, and the scope and spirit of the present disclosure are defined by the claims.

[0227] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0228] It is easy for those skilled in the art to understand that the above is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present application should be included in the scope of protection of the present application.

Claims

1. A three-dimensional surface morphology fast splicing method based on improved ORB, characterized in that: include: Based on the block matching method, a lateral sliding search is performed on the micro-nano surface structure to be spliced ​​to determine the overlapping area; In the overlapping area, an initial feature point set is extracted based on the FAST algorithm, non-maximum suppression is performed on the initial feature points, redundant feature points are removed, Harris corner point response values ​​of the feature points after non-maximum suppression are calculated, and screening is performed based on the response values ​​to obtain multiple feature points with the largest response values ​​to be matched; Describing the plurality of feature points to be matched with the largest response values ​​to obtain feature descriptors; Based on the feature descriptors of the feature points to be matched, the feature points to be matched are filtered using a K-nearest neighbor search algorithm to obtain coarse matching point pairs, and the coarse matching point pairs are finely matched using a decoupled RANSAC algorithm to obtain a three-dimensional rigid body transformation model; Based on the three-dimensional rigid body transformation model, micro-nano surface structure splicing and fusion are performed through transformation interpolation and overlapping area data fusion; The method of extracting an initial feature point set based on the FAST algorithm, performing non-maximum suppression processing on the initial feature points, and removing redundant feature points specifically includes: For each coordinate point in the overlapping area , whose height is , take this point as the center point, select the first number on the circle of the first radius m coordinate points ; Preset threshold t If there are a second number of consecutive coordinate points whose height difference with the center point is greater than the first threshold t ,Right now Then the coordinate point As candidate feature points; For candidate feature points , calculate its response value The candidate feature points Response value Compared with the response values ​​of all other candidate feature points within the range, if Response value is the local maximum, then keep is the initial feature point, otherwise it will be suppressed.

2. The method for rapid splicing of three-dimensional surface morphologies based on improved ORB according to claim 1, wherein: The calculating of the Harris corner point response value after the non-maximum suppression process, screening according to the response value, and obtaining multiple feature points to be matched with the largest response value specifically includes: Height map of the overlapping area , calculated by the Prewitt operator exist 、 Gradient in direction in, represents convolution; For each initial coordinate point, calculate the local autocorrelation matrix in, is a local window of size ; Calculate Harris corner response values in, is a matrix The determinant of is a matrix traces, is an empirical constant; Sort the calculated corner point response values ​​and retain multiple feature points to be matched with the largest response values.

3. The method for rapid splicing of three-dimensional surface morphology based on improved ORB according to claim 2, characterized in that: The step of describing the plurality of feature points to be matched with the largest response values ​​to obtain feature descriptors specifically includes: For the current feature point to be matched, determine a circle with a size of The local neighborhood of Generate in the local neighborhood For random coordinate point pairs ; For each pair of random coordinate points , compare their values ​​in the micro-nano surface structure height map and ,have All The comparison results of the coordinate point pairs are connected in sequence to form a length of The binary vector is the feature descriptor of the feature point 。 4. The method for rapid splicing of three-dimensional surface morphologies based on improved ORB according to claim 3, wherein: The method of filtering the feature points to be matched by using a K-nearest neighbor search algorithm based on the feature descriptors of the feature points to be matched to obtain rough matching point pairs specifically includes: For the set of feature points to be matched, the normalized Hamming distance is used to measure the similarity of the features. in, and It is a 256-bit feature descriptor; For the heightmap Feature points , in the height map Find the two nearest neighbor points of the feature points 、 , Feature points and in The two nearest neighbors in 、 ; Cross-validation and Is it a rough matching point pair, including verification and The nearest neighbors of are each other, the normalized Hamming distance of the nearest neighbor is less than the threshold, and the distance ratio between the nearest neighbor and the second nearest neighbor is less than the threshold, that is, in, 、 、 They are 、 、 feature descriptors.

5. The method for rapid splicing of three-dimensional surface morphology based on improved ORB according to claim 4, characterized in that: The method of performing fine matching on the rough matching point pairs by using a decoupled RANSAC algorithm to obtain a three-dimensional rigid body transformation model specifically includes: from Randomly select from the rough matching points , calculate the two-dimensional plane rigid body transformation model, use the model to map all point pairs, calculate the horizontal Euclidean distance as the reprojection error, if the reprojection error is less than the first preset threshold, then the point pair is considered as an inlier. Through multiple iterative calculations, the two-dimensional optimal inlier point set that satisfies horizontal alignment is obtained. The number of inliers is ; from Randomly select from two-dimensional optimal interior points Calculate the 3D planar rigid body transformation model, use the model to map all point pairs, calculate the Z-axis distance of corresponding point pairs as the reprojection error, and if the reprojection error is less than a second preset threshold, then consider the point pair as an inlier. Through multiple iterative calculations, obtain the 3D optimal inlier point set that satisfies spatial alignment; Based on the three-dimensional optimal interior point set, the three-dimensional rigid body transformation model is fitted by the least squares method. 。 6. The method for rapid splicing of three-dimensional surface morphology based on improved ORB according to claim 5, characterized in that: The micro-nano surface structure splicing and fusion is performed based on the three-dimensional rigid body transformation model through transformation interpolation and overlapping area data fusion, specifically including: Based on the bilinear interpolation principle, the Z-axis height of the interpolation point O is calculated as follows: in, 、 、 、 They are the height values ​​of the four mapping points around the point to be interpolated; After completing the interpolation and registration of the data, the overlapping areas are fused through weighted fusion. Weighted fusion assigns different weights to the two sets of data based on the distance between the feature points and the edge of the overlapping area. The weight of the overlapping area is: in, is the total width of the overlapping area, For data points Distance from the left border; Micro-nano surface structure data and Perform splicing and fusion, the splicing and fusion result is ,have 。 7. A three-dimensional surface morphology rapid splicing device based on improved ORB, characterized in that: include: A block matching module is used to perform a lateral sliding search on the micro-nano surface structure to be spliced ​​based on a block matching method to determine the overlapping area; A feature extraction module is used to extract an initial feature point set in the overlapping area based on the FAST algorithm, perform non-maximum suppression on the initial feature points, remove redundant feature points, calculate Harris corner point response values ​​of the feature points after the non-maximum suppression process, and screen them according to the response values ​​to obtain multiple feature points with the largest response values ​​to be matched; A descriptor acquisition module, configured to describe the plurality of feature points to be matched with the largest response values ​​and acquire feature descriptors; A matching module is used to filter the feature points to be matched using a K-nearest neighbor search algorithm based on the feature descriptors of the feature points to obtain coarse matching point pairs, and to perform fine matching on the coarse matching point pairs using a decoupled RANSAC algorithm to obtain a three-dimensional rigid body transformation model; A splicing and fusion module is used to perform splicing and fusion of micro-nano surface structures through transformation interpolation and overlapping area data fusion based on the three-dimensional space rigid body transformation model; The method of extracting an initial feature point set based on the FAST algorithm, performing non-maximum suppression processing on the initial feature points, and removing redundant feature points specifically includes: For each coordinate point in the overlapping area , whose height is , take this point as the center point, select the first number on the circle of the first radius m coordinate points ; Preset threshold t If there are a second number of consecutive coordinate points whose height difference with the center point is greater than the first threshold t ,Right now Then the coordinate point As candidate feature points; For candidate feature points , calculate its response value The candidate feature points Response value Compared with the response values ​​of all other candidate feature points within the range, if Response value is the local maximum, then keep is the initial feature point, otherwise it will be suppressed.

8. A three-dimensional surface morphology rapid splicing device based on improved ORB, characterized in that: The method comprises at least one processing unit and at least one storage unit, wherein the storage unit stores a computer program, and when the computer program is executed by the processing unit, the processing unit executes the steps of the method according to any one of claims 1 to 6.

9. A storage medium, characterized in that: It stores a computer program that can be executed by a three-dimensional surface morphology rapid splicing device based on an improved ORB. When the computer program runs on the three-dimensional surface morphology rapid splicing device based on an improved ORB, the three-dimensional surface morphology rapid splicing device based on an improved ORB executes the steps of the method described in any one of claims 1 to 6.

Images