A high-speed real-time line laser three-dimensional imaging method

Abstract

The application provides a high-speed real-time line laser three-dimensional imaging method, and relates to the technical field of data processing.The method comprises the following steps: extracting continuous strip units in basic strip data, forming strip columns according to the column direction of the continuous strip units, identifying the scanning angles of each strip column during imaging, calculating the consistency difference of the same strip column under adjacent scanning angles, calculating the mapping stability of the strip column under different angles to obtain an angular divergence value, calculating the strain degree generated by the strip column in space projection to obtain a sequence relaxation value, taking the sequence relaxation value as a stable correction factor, generating an angle window model according to the angle adjustment factor and the stable correction factor, determining the readable pixel range in each angle window, extracting strip image segments for recombination, calculating three-dimensional coordinates and forming three-dimensional strip coordinates, and obtaining real-time three-dimensional imaging data.The application can avoid the three-dimensional point cloud breakage phenomenon caused by strip mutation and realize line laser three-dimensional imaging.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a high-speed, real-time line laser three-dimensional imaging method. Background Technology

[0002] In the field of 3D vision measurement, line laser scanning technology based on laser triangulation is one of the mainstream methods for achieving high-speed, high-precision contour acquisition of object surfaces. The core of this technology lies in projecting a laser line onto the object surface using a line laser, forming a cross-section; a camera captures the deformed image of this laser line on the object surface from another angle; finally, based on the stable triangular geometric relationship formed between the laser, camera, and points on the object surface, the 3D coordinates of each point on the laser line are calculated. To obtain the 3D topography of the entire object surface, the cross-sectional contour needs to be accumulated. An advanced implementation method utilizes a rigid synchronous rotation structure, where the line laser and camera are fixed into a rigid projection and receiving module, and this module is synchronously rotated around an axis parallel to the laser line for scanning. During this process, the relative distance between the laser and the camera remains constant, therefore, for any measured point on the object surface, the corresponding triangulation relationship is constant.

[0003] However, even with this ideal rigid synchronous scanning structure, point cloud breaks can occur in high-speed, real-time applications due to abrupt changes in object geometry. For example, when scanning rapidly over objects with steep edges, deep grooves, or height abrupt changes such as the multi-step structure of a mobile phone frame, the laser line may be momentarily blocked or undergo drastic deformation. This can lead to significant jumps or discontinuities in the 3D reconstruction results of feature points on the same physical edge in consecutive scanning frames. The pixel position of the laser line in each frame may be stable, but the reconstructed 3D point cloud may show breaks or misalignments at geometric features. Summary of the Invention

[0004] The purpose of this invention is to provide a high-speed, real-time line laser three-dimensional imaging method, which aims to solve the problems mentioned in the background art.

[0005] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows:

[0006] A high-speed, real-time line laser three-dimensional imaging method, the method comprising:

[0007] Obtain the basic strip data reflected by linear light, extract continuous strip units from the basic strip data, form strip columns according to the column direction of the continuous strip units, and obtain the column index table;

[0008] Based on the column index table, the scanning angle of each strip column during imaging is identified, and the consistency difference of the same strip column in the three-dimensional mapping under adjacent scanning angles is calculated to obtain the angle offset matrix.

[0009] Based on the angle offset matrix, the spatial variation trend of the strip column in the angle scan is identified, the mapping stability of the strip column at different angles is calculated, and the angular convergence and divergence values ​​are obtained.

[0010] Based on the angular offset matrix, the overall deformation degree of the strip column in the angular scan is identified, the strain degree of the strip column in the spatial projection is calculated, and the sequence relaxation value is obtained.

[0011] By using the angular convergence and divergence values ​​as the angle adjustment factor and the sequence relaxation value as the stability correction factor, an angle window model is generated based on the angle adjustment factor and the stability correction factor to obtain dynamic segment data.

[0012] The basic strip data is remapped based on the dynamic segment data. The readable pixel range is determined within each angle window, and the strip image fragments are extracted and reconstructed to obtain the remapped strip data.

[0013] Based on the angle window index and column order index in the remapped strip data, the three-dimensional coordinates are calculated and three-dimensional strip coordinates are formed to obtain real-time three-dimensional imaging data.

[0014] Furthermore, based on the column index table, the scanning angle of each strip column during imaging is identified, and the consistency difference of the same strip column in the three-dimensional mapping under adjacent scanning angles is calculated to obtain the angle offset matrix, including:

[0015] By matching the corresponding scan record number to each strip column in the column index table according to the collection order, and combining the scan record number with the strip column, angle-related data is obtained;

[0016] Angle analysis data is obtained by grouping angle-related data according to scan record number and identifying the column order position of each strip column in different groups;

[0017] An angle difference sequence is obtained by arranging the column order positions of adjacent scan record numbers under the same band in the angle analysis data side by side and calculating the difference in the feature mapping consistency of the column order positions after the side by side.

[0018] By arranging the angle difference sequence into matrix rows according to the scan record number in ascending order, and arranging the corresponding strip column numbers into matrix columns in sequence, the angle offset matrix is ​​obtained.

[0019] Furthermore, based on the angle offset matrix, the spatial variation trend of the strip column during angular scanning is identified, the mapping stability of the strip column at different angles is calculated, and the angular convergence / divergence values ​​are obtained, including:

[0020] Based on the angle offset matrix, calculate the angle difference between adjacent scan record numbers to obtain the angle difference term; based on the angle offset matrix, calculate the feature mapping consistency change of the same strip column under adjacent scan angles to obtain the feature change term; based on the angle difference term and the feature change term, calculate the angular aggregation degree of the mapping trend of the strip column during the angle change process to obtain the aggregation degree term.

[0021] The characteristic change term is calculated by determining the intensity of characteristic changes in the strip column within each angular segment, resulting in the change scale term. The effective number of characteristic change terms in multiple scanning angles is used to calculate the effectiveness of the strip column, resulting in the effectiveness term. The aggregation degree term, change scale term, and effectiveness term are fused together to calculate the mapping stability of the strip column during angular scanning, resulting in the angular aggregation / dispersion value.

[0022] Furthermore, based on the angular offset matrix, the overall deformation degree of the strip column during angular scanning is identified, and the strain degree generated by the strip column in spatial projection is calculated to obtain the sequence relaxation value, including:

[0023] Based on the angle offset matrix, the feature mapping consistency difference change of the same strip column under consecutive scan record numbers is calculated to obtain the consistency difference change term; based on the consistency difference change term, the consistency difference changes corresponding to adjacent scan record numbers are sequentially compared to calculate the change factor between consistency difference changes to obtain the change factor term; based on the change factor term, the change factors of the same strip column under different scan angles are sequentially superimposed to calculate the continuous deformation path of the strip column in the angle domain to obtain the deformation path term.

[0024] Based on the deformation path term, the deformation amplitude of the strip column in each angle segment is statistically scaled to measure the overall scale of the deformation of the strip column during the entire angle scan, resulting in a deformation scale term. Based on the deformation scale term, the span of deformation of the strip column in multiple angle segments is calculated, resulting in a span quantity term. The deformation path term, deformation scale term, and span quantity term are fused together to calculate the strain degree of the strip column in the angle scan, resulting in the sequence relaxation value.

[0025] Furthermore, by using angular convergence / divergence values ​​as angle adjustment factors and sequence relaxation values ​​as stability correction factors, an angular window model is generated based on the angle adjustment factors and stability correction factors to obtain dynamic segment data, including:

[0026] By mapping the angular convergence and divergence values ​​to numerical values ​​used to adjust the range of scanning angle changes, the adjustment range of each angle interval is calculated to obtain the angle adjustment factor;

[0027] By converting the sequence relaxation value into a numerical value of segmentation fineness, the stability correction amount of the strip column in the angular domain is calculated, and the stability correction factor is obtained.

[0028] The starting point and span of the scanning angle are calculated based on the angle adjustment factor and the stability correction factor to determine the segment structure of the scanning angle and obtain the segment division data.

[0029] Based on the segment division data, the starting point and span of the segment are combined in the order of scanning angles to construct an angle window model and obtain dynamic segment data.

[0030] Furthermore, based on the segment division data, the segment starting point and segment span are combined according to the scanning angle order to construct an angle window model, obtaining dynamic segment data, including:

[0031] Based on the segment division data, the starting point and span of each segment are extracted, and the starting point of the segment is normalized to obtain the segment element data.

[0032] Based on the segment element data, the segment start point is paired with the corresponding segment span to obtain paired segments, and the angular coverage range of each paired segment is calculated to obtain window definition data.

[0033] Based on the window definition data, adjacent and overlapping angle coverage areas are merged and split, and the window number and angle coverage boundary are determined to obtain the angle window model;

[0034] Based on the angle window model, each window number is associated with the scan record number to determine the window number to which each scan record number belongs and its position range within the window, thus obtaining dynamic segment data.

[0035] Furthermore, the basic strip data is remapped based on the dynamic segment data. The readable pixel range is determined within each angular window, and strip image fragments are extracted and reconstructed to obtain remapped strip data, including:

[0036] Based on the dynamic segment data, the angle coverage of each window number is matched with the scan record number of the basic strip data to determine the basic strip data corresponding to each window number, thus obtaining the window strip data;

[0037] Based on the window strip data and column index table, the pixels within the angular coverage area of ​​each window number are filtered to determine the readable pixel range under each window number, thus obtaining the readable pixel data.

[0038] Based on the readable pixel data, adjacent and consecutive readable pixels under the same window number in both column and row order are aggregated to obtain a strip image pixel block set;

[0039] Based on the set of pixel blocks in the strip image, the strip image segments under the same window number are rearranged and stitched according to the scan record number order and column number order to determine the strip pixel sequence in each angle window and obtain the remapped strip data.

[0040] Furthermore, based on the readable pixel data, adjacent and consecutive readable pixels under the same window number in both column and row order are aggregated to obtain a striped image pixel block set, including:

[0041] Based on the readable pixel data, the readable pixels under the same window number are compared in column order and row direction to identify pixel association groups that are continuous in both dimensions, thus obtaining pixel association data.

[0042] Based on pixel association data, pixels that belong to the same column order direction but have discontinuities in row arrangement are linked across rows to fill in pixel gaps and obtain row completion data.

[0043] By integrating the row-direction pixels in the row-direction completion data with the pixels in the adjacent column order, a pixel combination that is continuously distributed in both column order and row order is formed, resulting in continuous pixel combination data.

[0044] Based on the continuous pixel combination data, the boundary range, column span, and row span of each pixel combination are summarized to determine the pixel block set of the local segment space of the strip, thus obtaining the pixel block set of the strip image.

[0045] Furthermore, based on the angle window index and column order index in the remapped strip data, three-dimensional coordinates are calculated and three-dimensional strip coordinates are formed to obtain real-time three-dimensional imaging data, including:

[0046] Based on the remapped strip data, the angle window index and column order index are combined to calculate the joint index position of each pixel block in the angle domain and column order domain, thus obtaining the joint index data;

[0047] Based on the combined index data, the angle index is converted into an angle displacement parameter, the column order index is converted into a column position parameter, and the two-dimensional position value of each pixel block is calculated to obtain the two-dimensional position data.

[0048] Based on the joint index data, the angular displacement parameter is used as the height derivation factor, and the column position parameter is used as the spatial expansion factor to calculate the spatial height value of each pixel block and obtain the height derivation data.

[0049] Based on the two-dimensional position data and height derivation data, the two-dimensional position value and spatial height value of each pixel block are integrated to determine the spatial coordinates of the strip pixels, thus obtaining the three-dimensional strip coordinates.

[0050] Furthermore, based on the two-dimensional position data and height derivation data, the two-dimensional position values ​​and spatial height values ​​of each pixel block are integrated to determine the spatial coordinates of the strip pixels, thus obtaining the three-dimensional strip coordinates, including:

[0051] Based on the two-dimensional position data, the column position parameters and row position parameters are classified to determine the planar position of each pixel block in the two-dimensional coordinate plane, thus obtaining the planar position data.

[0052] Based on the height derivation data, the height derivation values ​​corresponding to each plane position in the plane position data are matched, and the plane positions are mapped to the height derivation values ​​to obtain height matching data;

[0053] Based on the height matching data, the plane positions in the planar position are combined with their corresponding height derivation values ​​to form three-dimensional position units, thus obtaining position integration data;

[0054] Based on the location integration data, the three-dimensional location units are arranged according to the order of pixel blocks in the remapped strip data to determine the three-dimensional location sequence of strip pixels and obtain the three-dimensional strip coordinates.

[0055] The above-described solution of the present invention has at least the following beneficial effects:

[0056] This invention calculates the displacement difference between adjacent scanning angles for the same strip column and constructs an angle offset matrix with the scan record as the row and the strip column as the column, forming a two-dimensional displacement field in the discrete angle domain. This matrix provides a regular grid structure that can be directly used for difference, gradient, and convolution kernel, which facilitates first-order and second-order difference in the angle direction to characterize the drift trend and acceleration term. It makes the original arc-shaped drift trajectory explicit and binds it to the angle index, avoiding the need to associate strip positions through large-scale image search and eliminating index mismatch caused by row time-series jitter. This allows subsequent trend recognition, deformation measurement, and window construction to be completed in the same coordinate system, and the data path does not need to switch coordinate references.

[0057] This invention extracts the offset consistency of the strip column during the angular advancement process from the angle difference term and displacement change term, and combines it with the effective sample count to form the angular convergence and divergence value, thus obtaining a stability label. This label is an objective measure of the mapping volatility, and the result is a scalar sequence aligned with the discrete angle, which can be directly used as the index key for any subsequent segmentation, weighting, or resampling step. Since the stability measure, the strip column identity, and the scan record number exist simultaneously, subsequent processing does not need to perform texture correlation or template matching on the original image to determine the readable area. Data selection can be completed solely based on this scalar threshold decision, avoiding the introduction of new visual similarity ambiguities, and enabling the angular domain behavior of the strip to be rewritten into a queryable, segmentable, and indexable stability sequence.

[0058] This invention constructs a change factor by continuously scanning and recording consistent differences in changes, and then superimposes it along the angular sequence to obtain a deformation path term describing the non-rigid deformation trajectory of the same strip in the angular domain. At the same time, it statistically analyzes the deformation amplitude and angular span, and finally generates a sequence relaxation value. It provides a distinguishing degree for three typical behaviors: sudden jump, slow drift, and intermittent slip. It forms a joint metric for deformation intensity and cross-segment expansion for segmented modeling, so that any subsequent window-based pixel access can avoid the situation of mixing up the drastically deformed area with the stable area at the data level. The window boundary can be consistent with the deformation boundary, and the continuity of the strip in the angular domain and the possible break point location can be determined without accessing the original image.

[0059] This invention maps stability and deformation metrics to the start and span of angle segments, resulting in an angle window model bound to the scan record number. This model divides the angle domain into several defined, indexable segments and provides a range description for each segment where pixel access will occur. This ensures that all candidate domains of readable pixels are listed before resampling. The segmentation result can be directly converted into an ordered access plan by the sequence processor, avoiding uncertain branches introduced by temporarily searching for strip positions during frame-by-frame processing. At the same time, it ensures that any strip column has a clear data ownership relationship and spatial coverage boundary within any angle segment, facilitating deterministic coordinate transformation and splicing order in subsequent remapping.

[0060] This invention maps angular indices to angular displacement parameters and column order indices to column position parameters in a joint index plane, and then combines them with height derivation factors. This allows for the direct generation of spatial coordinates of strip pixels without backtracking to the original frame coordinates. Since each coordinate is bound to its source window number and scan record number, the three-dimensional coordinate sequence maintains a consistent order in the angular advancement direction. The spatial trajectory of the strip can be continuously output without additional time synchronization and inter-frame association, providing a monotonic input stream for backend surface fitting, strip-level point cloud stitching, and topology updating. This avoids height jumps and strip misconnections caused by index confusion, completing the deterministic transformation link of three-dimensional strip coordinates. Attached Figure Description

[0061] Figure 1 This is a flowchart of a high-speed, real-time line laser three-dimensional imaging method provided by an embodiment of the present invention. Detailed Implementation

[0062] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0063] like Figure 1 As shown, an embodiment of the present invention proposes a high-speed, real-time line laser three-dimensional imaging method, the method comprising:

[0064] Obtain the basic strip data reflected by linear light, extract continuous strip units from the basic strip data, form strip columns according to the column direction of the continuous strip units, and obtain the column index table;

[0065] Based on the column index table, the scanning angle of each strip column during imaging is identified, and the consistency difference of the same strip column in the three-dimensional mapping under adjacent scanning angles is calculated to obtain the angle offset matrix.

[0066] Based on the angle offset matrix, the spatial variation trend of the strip column in the angle scan is identified, the mapping stability of the strip column at different angles is calculated, and the angular convergence and divergence values ​​are obtained.

[0067] Based on the angular offset matrix, the overall deformation degree of the strip column in the angular scan is identified, the strain degree of the strip column in the spatial projection is calculated, and the sequence relaxation value is obtained.

[0068] By using the angular convergence and divergence values ​​as the angle adjustment factor and the sequence relaxation value as the stability correction factor, an angle window model is generated based on the angle adjustment factor and the stability correction factor to obtain dynamic segment data.

[0069] The basic strip data is remapped based on the dynamic segment data. The readable pixel range is determined within each angle window, and the strip image fragments are extracted and reconstructed to obtain the remapped strip data.

[0070] Based on the angle window index and column order index in the remapped strip data, the three-dimensional coordinates are calculated and three-dimensional strip coordinates are formed to obtain real-time three-dimensional imaging data.

[0071] In this embodiment of the invention, basic strip data of linear light reflection is acquired, and continuous strip units are extracted from the basic strip data. These continuous strip units are then arranged in column directions to form strip columns, resulting in a column order index table. The pixel set is confined to the connected regions related to the strips, while non-strip background, isolated bright spots, and speckles are excluded from the analysis domain. Based on the column order index table, the scanning angle of each strip column during imaging is identified, and the consistency difference of the same strip column on the 3D mapping under adjacent scanning angles is calculated to obtain an angle offset matrix. This allows subsequent trend identification, deformation estimation, and segmented modeling to be performed on the same grid. The process is performed within the matrix domain to avoid data alignment errors caused by switching between multiple coordinate systems. Based on the angle offset matrix, the spatial variation trend of the strip column in the angle scan is identified, the mapping stability of the strip column at different angles is calculated, and the angular convergence and divergence values ​​are obtained to avoid incorrect matching caused by similar appearances and ensure that the segment construction is completed entirely within the matrix domain. Based on the angle offset matrix, the overall deformation degree of the strip column in the angle scan is identified, the strain degree generated by the strip column in the spatial projection is calculated, and the sequence relaxation value is obtained so that the boundaries of subsequent segments can be aligned with the deformation boundaries, avoiding the confusion of drastic changes with stable areas.

[0072] By using angular convergence / divergence values ​​as angle adjustment factors and sequence relaxation values ​​as stabilization correction factors, an angle window model is generated based on these factors to obtain dynamic segment data. Each segment clearly defines its angular coverage boundary and expected access range, avoiding read window truncation at drastic changes and ensuring that subsequent pixel aggregation has complete context. The basic strip data is remapped based on the dynamic segment data. The readable pixel range is determined within each angle window, and strip image fragments are extracted and recombined to obtain remapped strip data. Subsequent operations can be completed within a fixed-size regular neighborhood, avoiding full-frame search and cross-window backtracking. Based on the angle window index and column order index in the remapped strip data, three-dimensional coordinates are calculated to form three-dimensional strip coordinates, resulting in real-time three-dimensional imaging data. The strips present a consistent point sequence in the angular advancement direction, facilitating downstream curve fitting and cross-strip stitching, reducing height jumps and strip misconnections caused by index mismatches, and meeting the real-time output requirements for latency and sequence.

[0073] This process involves acquiring basic strip data based on linear light reflection, extracting continuous stripe units from the basic stripe data, forming stripe columns according to the column direction, and obtaining a column index table. Specifically, this includes:

[0074] In the rigid synchronous rotational scanning system of this invention, a line laser and an image sensor are fixed into a rigid module and rotate around an axis for scanning. Based on this specific scenario, a laser line is first projected onto the object surface by the line laser in the rigid module. The image sensor synchronously acquires reflected light images at each scanning angle position, obtaining a raw image set arranged in a time sequence. Dark current elimination and brightness equalization preprocessing are performed on each frame of the raw image to suppress fixed-pattern noise and the influence of ambient light, resulting in a high-quality basic image frame sequence. This step utilizes the rigid rotation characteristics to acquire multi-view image data, providing a basis for observing the same object region from different angles.

[0075] For each frame of the base image, the laser stripes need to be accurately extracted and converted into structured data. Frequency domain filtering or background modeling techniques are used to extract low-frequency background components from the image. The high-frequency characteristics of the laser stripes are highlighted by the difference operation between the original image and the background image, suppressing background interference caused by uneven reflection from the object surface. The enhanced image is binarized using an adaptive thresholding algorithm to initially separate candidate regions for the laser stripes. Then, morphological opening operations are applied to remove small connected regions caused by speckle noise, preserving morphologically intact stripe regions. Connected regions in the binary image are labeled, with each connected region representing a stripe unit. For stripe breaks caused by occlusion or reflection attenuation, a stripe repair operation is performed based on the geometric proximity and grayscale similarity of adjacent pixels to obtain a continuous and complete set of stripe units.

[0076] Each continuous strip unit is subdivided along the image column direction. Within each image column, the pixels covered by that strip unit are aggregated to form a strip column. Simultaneously, the column position, row span, and grayscale statistics of this strip column in the current image are recorded. In this rigid system, because the camera and laser are relatively fixed, the column position of the strip column within a single frame is stable. This step provides a precise two-dimensional coordinate basis for subsequent triangulation calculations. After extracting the strip columns for all image frames corresponding to all scanning angles, the system constructs a column order index table. This table records the identification number, the column position and row span of the image frame to which it belongs, key associations such as the image frame number and corresponding scanning angle value, and the image acquisition timestamp for each strip column. This index table establishes an explicit association between strip columns and scanning angles, forming a searchable multi-view strip dataset, providing data support for subsequent analysis of the mapping consistency of the same object features under different scanning angles.

[0077] In a preferred embodiment of the present invention, the scanning angle of each strip column during imaging is identified according to the column order index table, and the consistency difference of the same strip column in the three-dimensional mapping under adjacent scanning angles is calculated to obtain the angle offset matrix, including:

[0078] By matching the corresponding scan record number to each strip column in the column index table according to the collection order, and combining the scan record number with the strip column, angle-related data is obtained;

[0079] Angle analysis data is obtained by grouping angle-related data according to scan record number and identifying the column order position of each strip column in different groups;

[0080] An angle difference sequence is obtained by arranging the column order positions of adjacent scan record numbers under the same band in the angle analysis data side by side and calculating the difference in the feature mapping consistency of the column order positions after the side by side.

[0081] By arranging the angle difference sequence into matrix rows according to the scan record number in ascending order, and arranging the corresponding strip column numbers into matrix columns in sequence, the angle offset matrix is ​​obtained.

[0082] In this embodiment of the invention, by matching the corresponding scan record numbers of each strip column in the column sequence index table according to the acquisition order, and combining the scan record numbers with the strip columns, angle-related data is obtained. This ensures that the behavior of the same strip column under different scanning angles can be tracked and identified, avoiding index loss due to missing or disordered column sequence information. By grouping the angle-related data according to the scan record numbers and identifying the column sequence position of each strip column in different groups, angle analysis data is obtained, avoiding re-performing image-level matching and strip detection on the original image. By arranging the column sequence positions of the same strip column under adjacent scan record numbers in the angle analysis data side by side, and calculating the difference in the feature mapping consistency of the arranged column sequence positions, an angle difference sequence is obtained. This reveals the lateral drift trend and amplitude fluctuation of the strip column as the angle changes, providing a numerical basis for the dynamic window model. By arranging the angle difference sequence into matrix rows according to the scan record numbers from smallest to largest, and arranging the corresponding strip column numbers into matrix columns in sequence, an angle offset matrix is ​​obtained, providing a complete and organized data foundation for subsequent construction of the angle window model and execution of dynamic remapping.

[0083] Specifically, angle analysis data is obtained by grouping angle-related data according to scan record numbers and identifying the column order of each strip in different groups, including:

[0084] First, the strip column numbers and their corresponding scan record numbers are extracted to form angle association data containing multiple combination pairs. Each combination pair represents the existence of a certain strip column in a certain frame of image, i.e., a specific scan record number. To achieve continuous position tracking of the strip column during multi-angle scanning, this angle association data needs to be logically grouped according to the scan record number. Specifically, the system sorts based on the scan record number and categorizes all combination pairs with the same scan number into the same angle group, ensuring that each angle group corresponds to a unique imaging frame. Subsequently, within each angle group, the system traverses all included strip column numbers and further calls the data recorded in the column order index table to extract the actual column order position of the strip column at the current angle, i.e., its horizontal pixel coordinates in the image frame. This is achieved through a table lookup method, where the system performs a key-value lookup for each strip column number and returns its column order value under the current scan record number. At this point, for each strip column number, a column order position associated with the scan record number is obtained. These column order position information are combined and saved according to the strip column number and scan record number to form the angle resolution data.

[0085] Specifically, by arranging the column order positions of adjacent scan record numbers under the same band in the angle analysis data side by side, and calculating the difference in the feature mapping consistency of the arranged column order positions, an angle difference sequence is obtained, which includes:

[0086] First, for each strip column number, the column position of that strip column under all scan record numbers is extracted and sorted in ascending order of scan record numbers to form a strip trajectory vector. This vector represents the lateral movement path of the strip column as the angle changes. For example, for strip column A, if it appears sequentially in scan record numbers 1, 2, and 3, with column positions x1, x2, and x3 respectively, the sequence [x1, x2, x3] is formed. The system performs a first-order difference calculation on this sequence to characterize the change in the column position of the strip column in the image between adjacent angles. The above calculation is implemented in a difference manner, that is, for any two adjacent values ​​xi and xi+1 in the sequence, the difference Δi = xi+1 − xi is calculated. For strip column A, the above operation will generate two differences: Δ1 = x2 − x1, Δ2 = x3 − x2, forming an angle difference sequence [Δ1, Δ2]. This sequence records the magnitude and direction of the lateral displacement of the strip column between consecutive angles, with positive and negative values ​​representing a tendency to drift left or right, respectively. The system performs the same operation on all strip column numbers, calculating the differences in their respective angle sequences, thus generating an angle difference sequence for each strip column. Finally, the system summarizes the angle difference sequences of all strip columns, arranging them in the order of scan record number as row and strip column number as column, to obtain the angle difference sequence.

[0087] In a preferred embodiment of the present invention, based on the angle offset matrix, the spatial variation trend of the strip column during angular scanning is identified, the mapping stability of the strip column at different angles is calculated, and the angular convergence / divergence value is obtained, including:

[0088] Based on the angle offset matrix, calculate the angle difference between adjacent scan record numbers to obtain the angle difference term; based on the angle offset matrix, calculate the feature mapping consistency change of the same strip column under adjacent scan angles to obtain the feature change term; based on the angle difference term and the feature change term, calculate the angular aggregation degree of the mapping trend of the strip column during the angle change process to obtain the aggregation degree term.

[0089] The characteristic change term is calculated by determining the intensity of characteristic changes in the strip column within each angular segment, resulting in the change scale term. The effective number of characteristic change terms in multiple scanning angles is used to calculate the effectiveness of the strip column, resulting in the effectiveness term. The aggregation degree term, change scale term, and effectiveness term are fused together to calculate the mapping stability of the strip column during angular scanning, resulting in the angular aggregation / dispersion value.

[0090] In this embodiment of the invention, the angle difference between adjacent scan record numbers is calculated based on the angle offset matrix to obtain the angle difference term, clarifying the quantitative relationship between the scan record number and the actual scan angle, thus laying the foundation for subsequent angle domain measurement. Based on the angle offset matrix, the feature mapping consistency change of the same strip column under adjacent scan angles is calculated to obtain the feature change term, enabling temporal dynamic monitoring of strip behavior and providing cross-frame tracking capability, directly reflecting strip stability. Based on the angle difference term and the feature change term, the angular aggregation degree of the mapping trend of the strip column during angle change is calculated to obtain the aggregation degree term, quantifying the offset consistency of the strip column in different angle segments. A smaller standard deviation indicates that the strip maintains a linear and stable trend along the angle scan path, contributing to the angular stability of the strip. The analysis provides real and measurable input; the feature change term calculates the intensity of feature changes in the strip column within each angle segment, resulting in a change scale term, which reveals the activity level of the strip column in the angular dimension. If the change scale term is large, it indicates that the strip column is significantly affected by angular disturbances; the effective number of feature change terms in multiple scanning angles calculates the effectiveness of the strip column, resulting in an effectiveness term, which can distinguish between long-term active strips and momentarily appearing interfering strips or erroneous trajectories, filtering noise input sources; the aggregation degree term, change scale term, and effectiveness term are fused to calculate the mapping stability of the strip column during angular scanning, obtaining the angular convergence and divergence value. This comprehensively considers the directional consistency, change amplitude, and overall participation activity of the strips during angular scanning, providing a key judgment benchmark for subsequent dynamic window generation.

[0091] In a preferred embodiment of the present invention, based on the angle offset matrix, the overall deformation degree of the strip column during angle scanning is identified, the strain degree generated by the strip column in spatial projection is calculated, and the sequence relaxation value is obtained, including:

[0092] Based on the angle offset matrix, the feature mapping consistency difference change of the same strip column under consecutive scan record numbers is calculated to obtain the consistency difference change term; based on the consistency difference change term, the consistency difference changes corresponding to adjacent scan record numbers are sequentially compared to calculate the change factor between consistency difference changes to obtain the change factor term; based on the change factor term, the change factors of the same strip column under different scan angles are sequentially superimposed to calculate the continuous deformation path of the strip column in the angle domain to obtain the deformation path term.

[0093] Based on the deformation path term, the deformation amplitude of the strip column in each angle segment is statistically scaled to measure the overall scale of the deformation of the strip column during the entire angle scan, resulting in a deformation scale term. Based on the deformation scale term, the span of deformation of the strip column in multiple angle segments is calculated, resulting in a span quantity term. The deformation path term, deformation scale term, and span quantity term are fused together to calculate the strain degree of the strip column in the angle scan, resulting in the sequence relaxation value.

[0094] In this embodiment of the invention, based on the angle offset matrix, the consistency difference change of the feature mapping of the same strip column under consecutive scan record numbers is calculated to obtain a consistency difference change term, capturing the dynamic trend of local bending or abrupt changes in the strip column during scanning; based on the consistency difference change term, the consistency difference changes corresponding to adjacent scan record numbers are sequentially compared to calculate the change factor between consistency difference changes, obtaining a change factor term, which evaluates the trend of strip position acceleration or deceleration during angle scanning, and is a key indicator for determining whether the strip is undergoing deformation focusing or diffusion; based on the change factor term, the change factors of the same strip column under different scanning angles are sequentially superimposed to calculate the continuous deformation path of the strip column in the angle domain, obtaining a deformation path term, analyzing whether the strip column is in a long-term deformation state, so that the deformation trend is no longer an isolated data point, but has an evolutionary law. The data sequence is analyzed. Based on the deformation path term, the deformation amplitude of the strip column in each angle segment is statistically scaled to measure the overall scale of deformation of the strip column during the entire angle scanning process, resulting in a deformation scale term. This effectively eliminates the influence of angle range and sampling density on the measurement results, making each strip column comparable in different angle domains. Based on the deformation scale term, the span of deformation of the strip column in multiple angle segments is calculated, resulting in a span quantity term, which expresses the extent to which the strip column is affected by geometric excitation during the scanning process and reflects the geographical distribution of deformation. The deformation path term, deformation scale term, and span quantity term are integrated to calculate the strain degree of the strip column in the angle scanning, resulting in a sequence relaxation value. This value reflects the elastic deformation capability of the strip column in the angle domain and is a comprehensive index that integrates trajectory intensity, amplitude range, and influence distribution, so that the dynamic window model no longer depends solely on local location.

[0095] In a preferred embodiment of the present invention, by using the angular convergence / divergence value as an angle adjustment factor and the sequence relaxation value as a stability correction factor, an angle window model is generated based on the angle adjustment factor and the stability correction factor to obtain dynamic segment data, including:

[0096] By mapping the angular convergence and divergence values ​​to numerical values ​​used to adjust the range of scanning angle changes, the adjustment range of each angle interval is calculated to obtain the angle adjustment factor;

[0097] By converting the sequence relaxation value into a numerical value of segmentation fineness, the stability correction amount of the strip column in the angular domain is calculated, and the stability correction factor is obtained.

[0098] The starting point and span of the scanning angle are calculated based on the angle adjustment factor and the stability correction factor to determine the segment structure of the scanning angle and obtain the segment division data.

[0099] Based on the segment division data, the starting point and span of the segment are combined in the order of scanning angles to construct an angle window model and obtain dynamic segment data.

[0100] In this embodiment of the invention, by mapping angular convergence values ​​to numerical values ​​used to adjust the range of scanning angle changes, the adjustment amplitude of each angle interval is calculated to obtain the angle adjustment factor. This achieves quantitative utilization of the stability of the strip column mapping in each angle segment, ensuring a balance between acquisition efficiency and integrity, and preventing redundant acquisition or loss of effective information due to excessively small windows. By converting sequence relaxation values ​​into numerical values ​​of segment division fineness, the stability correction amount of the strip column in the angle domain is calculated to obtain the stability correction factor. The granularity of the angle segment division is automatically adjusted according to the severity of strip deformation, increasing the sub-segment density in areas with large strain, reducing cross-window data jumps, and improving the coherence of subsequent splicing. Coordinate consistency; the starting point and span of the scanning angle are calculated based on the angle adjustment factor and the stability correction factor to determine the segment structure of the scanning angle and obtain segment division data. This ensures that the angle window model takes into account the changes in both strip stability and deformation trend, avoiding overly coarse or redundant divisions, and ensuring sufficient consistency and continuity of data within each window. Based on the segment division data, the starting point and span of the segment are combined in the order of the scanning angle to construct the angle window model, obtaining dynamic segment data. This provides a clear and unambiguous access structure for subsequent image remapping and 3D reconstruction. The reading of all strip image segments can be located in their respective angle windows through this model.

[0101] Specifically, by mapping angular convergence values ​​to numerical values ​​used to adjust the range of scanning angle changes, the adjustment amplitude of each angle interval is calculated to obtain the angle adjustment factor, which includes:

[0102] First, the angular convergence / dispersion values ​​for each strip column at different scanning angles are obtained. These values ​​reflect the mapping stability of the strips during angle changes. To convert the angular convergence / dispersion values ​​into values ​​for adjusting the scanning angle range, all angular convergence / dispersion values ​​are first normalized using linear scaling. The minimum value is set to zero, the maximum value to one, and the remaining values ​​are mapped proportionally to this interval. Then, a piecewise function is used to convert the normalized stability values ​​into adjustment coefficients. This piecewise function divides the normalized values ​​into multiple intervals. For example, the interval from 0 to 0.3 is defined as the high stability region, with the adjustment coefficient set to the base span; the interval from 0.3 to 0.7 is defined as the medium stability region, with the adjustment coefficient set to twice the base span; and the interval from 0.7 to 1 is defined as the low stability region, with the adjustment coefficient set to three times the base span. This method avoids small but unnecessary adjustment differences when the angular convergence / dispersion values ​​are close in value, improving the robustness of the model and enhancing the responsiveness of the adjustment coefficients in critical regions. All adjustment coefficients are recorded sequentially to form an adjustment coefficient sequence.

[0103] Specifically, by converting the sequence relaxation value into a numerical value indicating the fineness of segmentation, the stability correction amount of the strip column in the angular domain is calculated to obtain the stability correction factor, which includes:

[0104] First, the relaxation values ​​of the sequence reflect the strain trend of the strip column in the angular domain. A sliding window operation is performed on the relaxation value sequence, taking a fixed number of adjacent angular segments before and after each scanning angular point as the analysis range, and calculating the standard deviation and range of the relaxation values ​​within this window. The standard deviation is used to measure deformation fluctuation, and the range is used to assess the maximum change amplitude. The average of the standard deviation and range is used as the fluctuation measure of the current angular segment. Then, through a linear normalization method, this fluctuation measure is mapped to the interval between zero and one, with the minimum fluctuation mapped to zero and the maximum fluctuation mapped to one. To construct a stability correction factor, the normalized fluctuation value is input into a mapping function, which is a linear gain function. The input value is multiplied by an amplification factor greater than one to ensure a more obvious response in the high fluctuation region; for example, multiplying the normalized value by two is used as the correction factor. If the normalized fluctuation value of a certain angular segment is 0.5, then its correction factor is one. Finally, the correction factor is indexed with the angular number to form a stable correction factor sequence.

[0105] Specifically, the starting point and span of the scanning angle segment are calculated based on the angle adjustment factor and the stability correction factor to determine the segmentation structure of the scanning angle and obtain segmentation data, including:

[0106] After obtaining the adjustment coefficient sequence and the stabilization correction factor sequence, they are fused to generate segmentation data for scanning angles. The fusion method uses a weighted summation approach. First, weights are assigned to the adjustment coefficients and correction factors, for example, the adjustment coefficient weight is set to 0.6 and the correction factor weight to 0.4. Then, a weighted summation is performed on the adjustment coefficient and correction factor corresponding to each scanning angle number to obtain the fusion control factor. Subsequently, using the starting number of the scanning angle sequence as the initial point, a minimum window span is set as a baseline value, such as one or two degrees. This minimum span is multiplied by the fusion control factor to obtain the actual span of the current angle window. The current angle starting point plus this span value is used as the starting point of the next angle segment, and this process is repeated until the scanning angle sequence is completely covered. During window segmentation, if the span of a segment overlaps with the previous segment, their coverage areas are merged and the span is updated. If the span of a segment is too small, possibly due to both a small adjustment coefficient and a small correction factor, the system sets a minimum span threshold for forced adjustment to ensure that excessively dense segmentation between windows does not occur. Ultimately, all generated angle segments include the starting angle, span value, and segment number, forming segment division data.

[0107] In a preferred embodiment of the present invention, based on the segment division data, the segment starting point and segment span are combined in order of scanning angle to construct an angle window model, thereby obtaining dynamic segment data, including:

[0108] Based on the segment division data, the starting point and span of each segment are extracted, and the starting point of the segment is normalized to obtain the segment element data.

[0109] Based on the segment element data, the segment start point is paired with the corresponding segment span to obtain paired segments, and the angular coverage range of each paired segment is calculated to obtain window definition data.

[0110] Based on the window definition data, adjacent and overlapping angle coverage areas are merged and split, and the window number and angle coverage boundary are determined to obtain the angle window model;

[0111] Based on the angle window model, each window number is associated with the scan record number to determine the window number to which each scan record number belongs and its position range within the window, thus obtaining dynamic segment data.

[0112] In this embodiment of the invention, based on the segment division data, the starting point and span of each segment are extracted, and the starting point of the segment is normalized to obtain segment element data. This ensures that the angle window model can be described and stored on a unified angle scale, avoiding segment mismatch or window overlap caused by inconsistencies in coordinate systems between different data sources or imaging modes. Based on the segment element data, the starting point of the segment is paired with the corresponding span to obtain paired segments, and the angle coverage of each paired segment is calculated to obtain window definition data. This clearly defines the angle processing boundary of each window, avoiding situations where there are uncovered areas or overlapping conflicts between windows. The data is defined by merging and splitting adjacent and overlapping angular coverage areas, and determining the window number and angular coverage boundary to obtain the angular window model. This ensures the uniqueness and non-overlapping nature of the window boundaries, while allowing redundant window mechanisms for overlapping processing in some areas to avoid window cutting strip trajectories. Based on the angular window model, each window number is associated with the scan record number to determine the window number to which each scan record number belongs and its position range within the window, resulting in dynamic segment data. This allows for direct location of the position and affiliation of each frame image during subsequent image reconstruction and pixel extraction on a window-by-window basis, without the need for angle reverse lookup or sorting operations.

[0113] Specifically, based on the segment element data, the segment starting point is paired with the corresponding segment span to obtain paired segments, and the angular coverage range of each paired segment is calculated to obtain window definition data, which includes:

[0114] After analyzing the angular offset trend and calculating the deformation degree of the strip columns, multiple segmentation data items are calculated based on the extracted angular convergence and divergence values ​​and sequence relaxation values. Each item contains the strip column stability parameter and the recommended processing span at a certain scanning angle. To make these discrete segments operable, the starting angle or its corresponding scan record number in these segmentation data items is first structurally paired with the angle span, i.e., the number of consecutive frames or scan numbers recommended to be covered by the angle range. The system traverses each segmentation data item and directly combines the starting point number and the span value in the item into a tuple to form a paired segment structure. For example, if the starting angle of a strip column is number 85 and the recommended span is 7 frames, then the paired segment is (85,7), indicating that its angle coverage should include all angle samples from frame 85 to frame 91. After pairing, the system further performs coverage interval calculation for each paired segment. Specifically, it adds the segment's starting point and span to obtain the segment's ending angle number, i.e., the right boundary of the segment, thus deriving the complete angle coverage range. For example, the aforementioned (85,7) will be mapped to the coverage interval [85,92). All paired segments are sequentially transformed into a set of angle coverage ranges in a similar manner. Each coverage range is represented by two integers, namely, a half-open interval expression of the starting frame number and the ending frame number, forming a unified window definition data structure. At this time, the system assigns a temporary window identifier to each coverage range, such as Window1, Window2, etc. The identifier will serve as the primary key for referencing the window in subsequent processing. At the same time, each window records the starting point, ending point, and the original segment number from which its coverage angle originates.

[0115] Specifically, based on the window definition data, adjacent and overlapping angle coverage areas are merged and split, and the window number and angle coverage boundary are determined to obtain the angle window model, which includes:

[0116] After obtaining all window definition data, the system continues to execute the overlapping window processing logic. Since adjacent segments may have overlapping angle ranges in multiple strip columns or different angle segments—for example, if the previous window covers an angle range of [85, 92) and the next window covers [90, 97)—they overlap in the interval [90, 92). The system sorts all window definition data by starting point number in ascending order and compares the angle ranges of adjacent windows to see if they intersect. If an intersection exists, the system decides whether to merge or split the windows based on the set processing strategy. The merging strategy is suitable when strip columns have similar stability and deformation characteristics within the intersection interval. In this case, two windows can be merged into a new window, with the starting point being the smaller of the two values ​​and the ending point being the larger. The original two window numbers will be merged into a new window number, and the corresponding data mapping table will be updated accordingly. If two windows overlap angularly but exhibit significant differences in strip behavior (e.g., one segment deforms drastically while the other remains relatively stable), a splitting strategy is employed. This involves separating the overlapping area into an independent window segment, recording the boundaries of the preceding and following windows and the intermediate split window, and assigning a number to the new window. After completing all merging and splitting operations, the system constructs the final angle window model from all non-overlapping and ordered angle coverage intervals. Each window in this model has a unique number, a corresponding angle start and end boundary, and a correspondence with the original scan record number. The angle window model serves as the main control structure for strip image extraction and pixel remapping in the image processing workflow. It assigns a window number to each frame of scanned image and further serves as a data scheduling index for subsequent pixel filtering, strip stitching, and 3D calculation processes, ensuring accurate sampling and access boundaries in both the spatial and angle domains for each processing step. Through this processing, the system completes the mapping from segmented data to an operable angle window model, achieving standardized angle domain segmentation and controllable scheduling of strip scanned images.

[0117] In a preferred embodiment of the present invention, the basic strip data is remapped based on dynamic segment data, a readable pixel range is determined within each angle window, and strip image fragments are extracted and reconstructed to obtain remapped strip data, including:

[0118] Based on the dynamic segment data, the angle coverage of each window number is matched with the scan record number of the basic strip data to determine the basic strip data corresponding to each window number, thus obtaining the window strip data;

[0119] Based on the window strip data and column index table, the pixels within the angular coverage area of ​​each window number are filtered to determine the readable pixel range under each window number, thus obtaining the readable pixel data.

[0120] Based on the readable pixel data, adjacent and consecutive readable pixels under the same window number in both column and row order are aggregated to obtain a strip image pixel block set;

[0121] Based on the set of pixel blocks in the strip image, the strip image segments under the same window number are rearranged and stitched according to the scan record number order and column number order to determine the strip pixel sequence in each angle window and obtain the remapped strip data.

[0122] In this embodiment of the invention, based on dynamic segment data, the angular coverage range of each window number is matched with the scan record number of the basic strip data to determine the basic strip data corresponding to each window number, thus obtaining window strip data. This achieves the classification and organization of strip data in the window space, enabling subsequent processing to focus on the local data within each window. Based on the window strip data and column index table, pixels within the angular coverage range of each window number are filtered to determine the readable pixel range under each window number, obtaining readable pixel data. Pixel data in the basic image that does not belong to the effective angular window range is removed, eliminating background and errors. Invalid pixels in reflective and non-striped areas; based on readable pixel data, adjacent and consecutive readable pixels in the same window number are aggregated in column and row order to obtain a striped image pixel block set, which solves the problem of short-term breakage of the strip due to uneven reflection or occlusion, and provides continuity support for the reconstruction of the striped image; based on the striped image pixel block set, the striped image segments in the same window number are rearranged and stitched according to the scan record number order and column number order to determine the striped pixel sequence in each angle window, and obtain remapped striped data, realizing the mapping of the striped image from the irregular curved position in the original image frame to a unified coordinate plane.

[0123] Specifically, based on the window strip data and column index table, pixels within the angular coverage area of ​​each window number are filtered to determine the readable pixel range for each window number, resulting in readable pixel data, including:

[0124] First, pixel filtering of the basic strip data is performed based on the angle window number recorded in the dynamic segment data and the angle range it covers. This process relies on the joint drive of the window strip data and the column order index table, and the specific operations are as follows: For each angle window number, the system first retrieves its corresponding angle coverage range, such as the scan record number range from frame_m to frame_n. Then, it extracts all strip pixel information belonging to the scan number range from the basic strip data, forming the original strip frame set under that window number. Each strip frame corresponds to the bright stripe formed by the linear laser projection in an image frame, which contains multiple strip units. These strip units have been assigned unique column order numbers through the column order index table. In order to filter out pixels that meet the readability conditions from these image frames, the column order index needs to be mapped one-to-one with the two-dimensional coordinates of the strip in the image frame. The system will traverse all image frames under the current window number, extract pixel blocks for each strip unit, and bind its column order number to the scan number of the frame. The system determines which stripe units exist consecutively in the image plane by accessing the column index table, such as having a certain degree of adjacency in the row and column directions. Based on a set connectivity threshold, such as a consecutive pixel count greater than 3 and a grayscale value exceeding a set brightness threshold, these units are marked as readable pixel units. All pixel positions (x, y), their column index, frame number, and pixel grayscale values ​​that meet these conditions are recorded in a unified pixel data table, which constitutes the readable pixel data for the current window number.

[0125] Specifically, based on the pixel block set of the strip image, the strip image segments under the same window number are rearranged and stitched according to the scan record number order and column number order to determine the strip pixel sequence within each angle window, thus obtaining the remapped strip data, including:

[0126] First, the system groups the readable pixel data according to the scan record number, ensuring that the data under the same window number are arranged in ascending order over time. Within each time group, the system further sorts the strip pixel blocks according to the column sequence number, ensuring that the order of the stripe columns matches the order in which the stripes sweep across the object in the actual imaging. After sorting, the system allocates an image buffer in memory that matches the stripe region to hold the remapped image. Subsequently, for each group of pixel blocks, the system maps them into a two-dimensional rearrangement matrix. The horizontal axis of this matrix represents the column sequence number, and the vertical axis represents the scan record number. Each element in the matrix corresponds to a group of consecutive pixel blocks in the original image. Within each matrix cell, the system fills in the corresponding pixel block to maintain the geometric relationship of the original stripe segment. For stripe interruptions caused by occlusion or reflection, the system inserts empty pixels or zero values ​​as padding to ensure the rectangular integrity of the overall structure. After the rearrangement is completed, the system reassembles the pixel sequence in the matrix to generate a strip pixel sequence within the current angle window. This strip pixel sequence has a clear structure and is arranged in a regular manner. In the end, each window number corresponds to an independent remapped strip data.

[0127] In a preferred embodiment of the present invention, based on readable pixel data, adjacent and consecutive readable pixels under the same window number in both column and row order are aggregated to obtain a striped image pixel block set, including:

[0128] Based on the readable pixel data, the readable pixels under the same window number are compared in column order and row direction to identify pixel association groups that are continuous in both dimensions, thus obtaining pixel association data.

[0129] Based on pixel association data, pixels that belong to the same column order direction but have discontinuities in row arrangement are linked across rows to fill in pixel gaps and obtain row completion data.

[0130] By integrating the row-direction pixels in the row-direction completion data with the pixels in the adjacent column order, a pixel combination that is continuously distributed in both column order and row order is formed, resulting in continuous pixel combination data.

[0131] Based on the continuous pixel combination data, the boundary range, column span, and row span of each pixel combination are summarized to determine the pixel block set of the local segment space of the strip, thus obtaining the pixel block set of the strip image.

[0132] In this embodiment of the invention, based on readable pixel data, readable pixels under the same window number are compared in both column and row dimensions to identify pixel association groups that are continuous in both dimensions, thus obtaining pixel association data. This allows for the accurate extraction of spatially connected strip structure segments from a large range of readable pixels, avoiding the incorrect classification of isolated noise, background bright spots, or boundary pseudo-pixels into the strip structure. Based on the pixel association data, pixels belonging to the same column direction but with discontinuous row arrangement are linked across rows to fill in pixel gaps, obtaining row completion data. This repairs strip breaks caused by differences in material reflection or sensor instantaneous sampling omissions, improving the integrity of the strip structure in the row dimension. This provides a continuous foundation for pixel aggregation and ridge fitting in subsequent strip reconstruction. By integrating the row pixels in the row completion data with the pixels in their adjacent columns, a pixel combination continuously distributed in both column and row directions is formed, resulting in continuous pixel combination data. This compensates for the jagged breaks in the strips caused by slope changes, making the overall structure of the strip data more regular. Based on the continuous pixel combination data, the boundary range, column span, and row span of each pixel combination are summarized to determine the pixel block set of the local segment space of the strip, resulting in a strip image pixel block set. Each block corresponds to a local scan result with physical strip significance, providing a directly usable minimum reconstruction unit for 3D data mapping and reconstructed surface generation.

[0133] Specifically, based on readable pixel data, readable pixels under the same window number are compared in both column and row dimensions to identify pixel association groups that are continuous in both dimensions, thus obtaining pixel association data, which includes:

[0134] First, for each window number, extract its corresponding set of all readable pixels. Each pixel in this set has a distinct column number and row number, forming a two-dimensional coordinate set. During implementation, an adjacency graph construction method is used to mesh all pixels according to their column number (c) and row direction (r). In this mapping relationship, starting from any pixel, iterate through its eight neighboring pixels (top, bottom, left, right, and four diagonals) to see if they exist in the readable pixel set. If they exist and their grayscale values ​​are close (difference less than 15), they are considered spatially continuous and included in the same connected group. A depth-first search is used to construct connected regions starting from any unprocessed pixel until all readable pixels have been processed. Finally, each identified set of connected pixels represents a pixel association group that is continuous in both column and row dimensions.

[0135] Specifically, based on pixel association data, pixels belonging to the same column order direction but with discontinuities in row arrangement are linked across rows to fill in pixel gaps, resulting in row completion data, which includes:

[0136] First, for each pixel association group, it is categorized and aggregated according to column number to obtain pixel subsequences under each column number. Then, for the row array under each column number, difference calculation is performed to identify whether there is a jump interval between row numbers, that is, whether the difference between the row numbers of two adjacent pixels is greater than 1. If the difference is between 1 and 3 pixels, and the grayscale difference of the pixels before and after the jump is less than a set threshold, it is determined that there is a strip break in the gap area, and a completion operation can be performed. Pixels in the middle row are inserted in the broken strip interval. The grayscale value can be the average grayscale value of the pixels at both ends of the broken strip, and the position coordinates are consistent with the column order before and after. The completed pixels need to be labeled as interpolated to distinguish them from real captured pixels. After completion, the pixel group structure is updated to make it continuous in the row direction, thereby restoring the broken strip area caused by occlusion, reflection attenuation, or image acquisition abnormalities, forming a structurally complete strip segment, and obtaining row-direction completed data.

[0137] Specifically, by integrating the row-direction pixels in the row-direction completion data with the pixels in their adjacent column order, a pixel combination continuously distributed in both column and row order is formed, resulting in continuous pixel combination data, which specifically includes:

[0138] By performing column-direction integration on the row-direction completion data, the continuity of the strip structure is further restored in both column order and row direction. First, each pixel in the pixel group with completed strip filling is traversed one by one, checking for valid pixels with the same row direction number in both column order +1 and column order -1 directions. If found, the current pixel is combined with pixels in the adjacent column to form a horizontal combination unit. This process is based on the principle of local symmetry in the column direction, prioritizing the integration of pixel blocks with consistent row direction and continuous column order. To prevent erroneous merging, it is necessary to verify whether the grayscale difference between adjacent pixels is within a set range (e.g., no more than 20), and to determine whether the column order span and row direction span of the combined block are within the expected strip width range (e.g., column order 320, row direction 550). All combination units that meet the requirements are marked as valid continuous pixel blocks and uniformly stored in the continuous pixel combination data.

[0139] In a preferred embodiment of the present invention, three-dimensional coordinates are calculated and three-dimensional strip coordinates are formed based on the angle window index and column order index in the remapped strip data to obtain real-time three-dimensional imaging data, including:

[0140] Based on the remapped strip data, the angle window index and column order index are combined to calculate the joint index position of each pixel block in the angle domain and column order domain, thus obtaining the joint index data;

[0141] Based on the combined index data, the angle index is converted into an angle displacement parameter, the column order index is converted into a column position parameter, and the two-dimensional position value of each pixel block is calculated to obtain the two-dimensional position data.

[0142] Based on the joint index data, the angular displacement parameter is used as the height derivation factor, and the column position parameter is used as the spatial expansion factor to calculate the spatial height value of each pixel block and obtain the height derivation data.

[0143] Based on the two-dimensional position data and height derivation data, the two-dimensional position value and spatial height value of each pixel block are integrated to determine the spatial coordinates of the strip pixels, thus obtaining the three-dimensional strip coordinates.

[0144] In this embodiment of the invention, the angle window index and column order index are combined based on the remapped strip data to calculate the joint index position of each pixel block in the angle domain and column order domain, thus obtaining joint index data. This ensures that under multi-angle scanning conditions, the strip data will not experience index confusion or positional ambiguity due to pixel drift, effectively avoiding the problem of discontinuous strip seams during 3D point cloud construction. Based on the joint index data, the angle index is converted into an angle displacement parameter, and the column order index is converted into a column position parameter. The two-dimensional position value of each pixel block is calculated to obtain two-dimensional position data. The two-dimensional image index is then converted into planar coordinates in the physical world, constructing a coordinate mapping relationship from the image domain to physical space, enabling... Each pixel block possesses precise two-dimensional positional semantics. Based on the joint index data, the angular displacement parameter is used as the height derivation factor, and the column position parameter is used as the spatial expansion factor to calculate the spatial height value of each pixel block, obtaining height derivation data. This ensures that the accuracy of height estimation remains even when the scanning angle changes, effectively eliminating height compression or expansion distortion caused by the angle. Based on the two-dimensional positional data and the height derivation data, the two-dimensional positional value and spatial height value of each pixel block are integrated to determine the spatial coordinates of the strip pixels, obtaining three-dimensional strip coordinates. This ensures that the accuracy of height estimation remains even when the scanning angle changes, effectively eliminating height compression or expansion distortion caused by the angle.

[0145] Specifically, based on the remapped strip data, the angle window index and column order index are combined to calculate the joint index position of each pixel block in the angle domain and column order domain, thus obtaining the joint index data, which includes:

[0146] First, for the strip image data after remapping, the system scans each strip pixel block row by row and column by column, extracting two core index information of the pixel block in the remapping structure: its angle window number and its column sequence number in the strip's column direction. These two types of indices are combined as two-dimensional discrete coordinates, where the angle window number represents the scanning angle segment corresponding to the current pixel, and the column sequence number represents the horizontal arrangement position of the strip in the optically sectioned image. Specifically, the system constructs a set of joint index tables, which number each pixel block in the form of [AW_i, CS_j], where AW represents the angle window number, CS represents the column sequence number, and i and j are the pixel block indices. Simultaneously, a mapping table is established between this joint number and the actual storage location of the pixel block within the image. The system also performs normalization processing on this joint index to ensure the uniqueness and continuity of the numbers in multi-threaded or parallel computing scenarios, avoiding the generation of duplicate entries or invalid indexes, resulting in joint index data.

[0147] Specifically, based on the combined index data, the angular index is converted into an angular displacement parameter, and the column order index is converted into a column position parameter. The two-dimensional position value of each pixel block is calculated to obtain the two-dimensional position data, which includes:

[0148] First, the angle window number and column sequence number recorded in the composite index table are read, and then converted into position parameters in real physical space through table lookup and geometric transformation. The angle window number serves as the angle index, and is then used by a system-preset calibration function, such as... AW_i is mapped to the angular displacement value on the rotating scan platform. Indicates the angular spacing between adjacent angular windows. This is the initial scan angle. In specific calculations, this angle value can be further multiplied by the rotation baseline length R to obtain the actual displacement on the scan path, such as the offset in the Y direction. Meanwhile, the column number CS_j is determined by the inter-column pixel spacing. relative to the reference point position Perform linear transformations, for example The system obtains the physical coordinates of the pixel block in the X direction. Combining the angular displacement and column position, the system constructs a two-dimensional physical position coordinate [X,Y]. Y is usually derived from the angular displacement through circular arc unfolding or planar projection. The two-dimensional position data of all pixel blocks are uniformly stored in the two-dimensional position data.

[0149] Specifically, based on the joint index data, the angular displacement parameter is used as the height derivation factor, and the column position parameter is used as the spatial expansion factor to calculate the spatial height value of each pixel block, thus obtaining the height derivation data, which includes:

[0150] Based on the angular displacement parameters and column position parameters obtained in the second step, the system derives the height value, i.e., the Z-coordinate, using the imaging geometry model of laser optical sectioning. In this process, the system first... The reflected light is converted into a corresponding laser projection direction vector, and the relative position of this vector with the sensor receiving plane is used to calculate the projection point of the reflected light in three-dimensional space; simultaneously, the system will adjust the column position... As the point where the projection line intersects the receiving surface, the height value perpendicular to the light source baseline is determined by the trigonometric relationship formed by this value and the angular direction. In specific implementations, the following geometric relationship model may be used: ,in Indicates the baseline length between the sensor and the laser. This represents the focal length of the system or the distance between the projection plane and the reference plane. This represents the column-oriented position value. The system performs the above calculation for each pixel block to obtain its derived height value in three-dimensional space. The system then writes this value as a spatial attribute of the pixel block into the height data table. To improve accuracy, the system can also perform angular nonlinear calibration, distortion correction, and error regression correction before calculation. Finally, the system will calculate the two-dimensional position [ , [and the derived value of this height] Binding is performed to ensure that each pixel block has a complete, accurate, and continuous spatial coordinate description in three-dimensional space.

[0151] In a preferred embodiment of the present invention, based on two-dimensional position data and height derivation data, the two-dimensional position value and spatial height value of each pixel block are integrated to determine the spatial coordinates of the strip pixels, thereby obtaining the three-dimensional strip coordinates, including:

[0152] Based on the two-dimensional position data, the column position parameters and row position parameters are classified to determine the planar position of each pixel block in the two-dimensional coordinate plane, thus obtaining the planar position data.

[0153] Based on the height derivation data, the height derivation values ​​corresponding to each plane position in the plane position data are matched, and the plane positions are mapped to the height derivation values ​​to obtain height matching data;

[0154] Based on the height matching data, the plane positions in the planar position are combined with their corresponding height derivation values ​​to form three-dimensional position units, thus obtaining position integration data;

[0155] Based on the location integration data, the three-dimensional location units are arranged according to the order of pixel blocks in the remapped strip data to determine the three-dimensional location sequence of strip pixels and obtain the three-dimensional strip coordinates.

[0156] In this embodiment of the invention, based on two-dimensional position data, column-direction position parameters and row-direction position parameters are categorized to determine the planar position of each pixel block in the two-dimensional coordinate plane, thus obtaining planar position data. This ensures that all pixels can be directly located and compared at the position level, providing a stable planar positioning basis for subsequent height assignment and 3D stitching, and avoiding horizontal and vertical misalignment problems caused by angle switching. Based on height derivation data, the height derivation values ​​corresponding to each planar position in the planar position data are matched, and the planar positions are mapped to the height derivation values ​​to obtain height matching data. This achieves the mapping from two-dimensional planar positions to the height dimension in three-dimensional spatial coordinates, constructing a three-dimensional point... One of the fundamental elements is the Z-axis coordinate. Based on the height matching data, the planar positions in the planar position are combined with their corresponding height derivation values ​​to form three-dimensional position units, resulting in position integration data. This completes the conversion from two-dimensional image pixels to three-dimensional physical space points, realizing the spatial representation of strip pixels. Based on the position integration data, the three-dimensional position units are arranged according to the order of pixel blocks in the remapped strip data to determine the three-dimensional position sequence of strip pixels, obtaining three-dimensional strip coordinates. This realizes the conversion of the strip from a discrete set of points to an ordered geometric sequence in three-dimensional space. The three-dimensional strip coordinates have continuity and trajectory consistency, ensuring the consistency and correctness of the point sequence direction in three-dimensional surface reconstruction.

[0157] Specifically, based on the height matching data, each planar position in the planar position is combined with its corresponding derived height value to form a three-dimensional position unit, resulting in integrated position data, which includes:

[0158] After establishing a one-to-one matching relationship between the planar position data and the corresponding derived height values, these data need to be fused to generate spatial coordinate information. Specifically, based on the height matching data, the two-dimensional planar position (i.e., the column position parameter and row position parameter) in each record is combined with its corresponding derived height value, that is, the column position value in the planar position is extracted separately. Row position value And the height value derived from the angle. By combining the three elements into a single three-dimensional coordinate unit This unit represents the physical location of a certain remapped strip pixel block in space. The combination process can be implemented through a data structure mapping table. For example, using a combined index (i.e., angular window index and column order index) as the key and three-dimensional coordinates as the value, all pixel spatial locations can be uniformly organized to obtain integrated data containing the spatial locations of all strip pixel points.

[0159] Specifically, based on the location integration data, the three-dimensional location units are arranged according to the order of pixel blocks in the remapped strip data to determine the three-dimensional location sequence of the strip pixels, thereby obtaining the three-dimensional strip coordinates. This process includes:

[0160] To further develop a three-dimensional representation of the strip structure, these three-dimensional positional units need to be reordered according to the arrangement order of the original strip image fragments in the remapped strip data, generating a three-dimensional positional sequence with spatial continuity and sequence consistency. First, the arrangement order information of each pixel block in the remapped strip data is read. This information can be obtained from the sorting key composed of the scan record number and the column sequence number. For example, let the scan record number be... The column number is Then they are combined to form a sort key. The logical order of the strip image segments can be obtained by sorting them in ascending order by key value. Then, the corresponding 3D position units are extracted sequentially according to this order. A three-dimensional coordinate sequence unfolding according to the scanning process is constructed. This sequence maintains the spatial continuity of the strips as the scanning angle advances during the physical imaging process, ensuring that the spatial arrangement of each strip pixel corresponds one-to-one with its acquisition process in the image domain, ultimately forming complete and ordered three-dimensional strip coordinate data.

[0161] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A high-speed, real-time line laser three-dimensional imaging method, characterized in that, The method includes: Obtain the basic strip data reflected by linear light, extract continuous strip units from the basic strip data, form strip columns according to the column direction of the continuous strip units, and obtain the column index table; Based on the column index table, the scanning angle of each strip column during imaging is identified, and the consistency difference of the same strip column in the three-dimensional mapping under adjacent scanning angles is calculated to obtain the angle offset matrix. Based on the angle offset matrix, the spatial variation trend of the strip column in the angle scan is identified, the mapping stability of the strip column at different angles is calculated, and the angular convergence and divergence values ​​are obtained. Based on the angular offset matrix, the overall deformation degree of the strip column in the angular scan is identified, the strain degree of the strip column in the spatial projection is calculated, and the sequence relaxation value is obtained. By using the angular convergence and divergence values ​​as the angle adjustment factor and the sequence relaxation value as the stability correction factor, an angle window model is generated based on the angle adjustment factor and the stability correction factor to obtain dynamic segment data. The basic strip data is remapped based on the dynamic segment data. The readable pixel range is determined within each angle window, and the strip image fragments are extracted and reconstructed to obtain the remapped strip data. Based on the angle window index and column order index in the remapped strip data, the three-dimensional coordinates are calculated and three-dimensional strip coordinates are formed to obtain real-time three-dimensional imaging data. Based on the angular offset matrix, the overall deformation degree of the strip column during angular scanning is identified, and the strain degree generated by the strip column in spatial projection is calculated to obtain the sequence relaxation value, including: Based on the angle offset matrix, the feature mapping consistency difference change of the same strip column under consecutive scan record numbers is calculated to obtain the consistency difference change term; based on the consistency difference change term, the consistency difference changes corresponding to adjacent scan record numbers are sequentially compared to calculate the change factor between consistency difference changes to obtain the change factor term; based on the change factor term, the change factors of the same strip column under different scan angles are sequentially superimposed to calculate the continuous deformation path of the strip column in the angle domain to obtain the deformation path term. Based on the deformation path term, the deformation amplitude of the strip column in each angle segment is statistically scaled to measure the overall scale of the deformation of the strip column during the entire angle scan, resulting in a deformation scale term. Based on the deformation scale term, the span of deformation of the strip column in multiple angle segments is calculated, resulting in a span quantity term. The deformation path term, deformation scale term, and span quantity term are fused together to calculate the strain degree of the strip column in the angle scan, resulting in the sequence relaxation value.

2. The high-speed real-time line laser three-dimensional imaging method according to claim 1, characterized in that, Based on the column index table, the scanning angle of each strip column during imaging is identified, and the consistency difference of the same strip column in the 3D mapping under adjacent scanning angles is calculated to obtain the angle offset matrix, including: By matching the corresponding scan record number to each strip column in the column index table according to the collection order, and combining the scan record number with the strip column, angle-related data is obtained; Angle analysis data is obtained by grouping angle-related data according to scan record number and identifying the column order position of each strip column in different groups; An angle difference sequence is obtained by arranging the column order positions of adjacent scan record numbers under the same band in the angle analysis data side by side and calculating the difference in the feature mapping consistency of the column order positions after the side by side. By arranging the angle difference sequence into matrix rows according to the scan record number in ascending order, and arranging the corresponding strip column numbers into matrix columns in sequence, the angle offset matrix is ​​obtained.

3. The high-speed real-time line laser three-dimensional imaging method according to claim 2, characterized in that, Based on the angular offset matrix, the spatial variation trend of the strip column during angular scanning is identified, the mapping stability of the strip column at different angles is calculated, and the angular convergence / divergence values ​​are obtained, including: Based on the angle offset matrix, calculate the angle difference between adjacent scan record numbers to obtain the angle difference term; based on the angle offset matrix, calculate the feature mapping consistency change of the same strip column under adjacent scan angles to obtain the feature change term; based on the angle difference term and the feature change term, calculate the angular aggregation degree of the mapping trend of the strip column during the angle change process to obtain the aggregation degree term. The characteristic change term is calculated by determining the intensity of characteristic changes in the strip column within each angular segment, resulting in the change scale term. The effective number of characteristic change terms in multiple scanning angles is used to calculate the effectiveness of the strip column, resulting in the effectiveness term. The aggregation degree term, change scale term, and effectiveness term are fused together to calculate the mapping stability of the strip column during angular scanning, resulting in the angular aggregation / dispersion value.

4. The high-speed real-time line laser three-dimensional imaging method according to claim 3, characterized in that, By using angular convergence / disconvergence values ​​as angle adjustment factors and sequence relaxation values ​​as stability correction factors, an angular window model is generated based on the angle adjustment factors and stability correction factors to obtain dynamic segment data, including: By mapping the angular convergence and divergence values ​​to numerical values ​​used to adjust the range of scanning angle changes, the adjustment range of each angle interval is calculated to obtain the angle adjustment factor; By converting the sequence relaxation value into a numerical value of segmentation fineness, the stability correction amount of the strip column in the angular domain is calculated, and the stability correction factor is obtained. The starting point and span of the scanning angle are calculated based on the angle adjustment factor and the stability correction factor to determine the segment structure of the scanning angle and obtain the segment division data. Based on the segment division data, the starting point and span of the segment are combined in the order of scanning angles to construct an angle window model and obtain dynamic segment data.

5. The high-speed real-time line laser three-dimensional imaging method according to claim 4, characterized in that, Based on the segment division data, the segment start point and segment span are combined according to the scanning angle order to construct an angle window model, thereby obtaining dynamic segment data, including: Based on the segment division data, the starting point and span of each segment are extracted, and the starting point of the segment is normalized to obtain the segment element data. Based on the segment element data, the segment start point is paired with the corresponding segment span to obtain paired segments, and the angular coverage range of each paired segment is calculated to obtain window definition data. Based on the window definition data, adjacent and overlapping angle coverage areas are merged and split, and the window number and angle coverage boundary are determined to obtain the angle window model; Based on the angle window model, each window number is associated with the scan record number to determine the window number to which each scan record number belongs and its position range within the window, thus obtaining dynamic segment data.

6. The high-speed real-time line laser three-dimensional imaging method according to claim 5, characterized in that, The base strip data is remapped based on dynamic segment data. Within each angular window, the readable pixel range is determined, and strip image fragments are extracted and reconstructed to obtain remapped strip data, including: Based on the dynamic segment data, the angle coverage of each window number is matched with the scan record number of the basic strip data to determine the basic strip data corresponding to each window number, thus obtaining the window strip data; Based on the window strip data and column index table, the pixels within the angular coverage area of ​​each window number are filtered to determine the readable pixel range under each window number, thus obtaining the readable pixel data. Based on the readable pixel data, adjacent and consecutive readable pixels under the same window number in both column and row order are aggregated to obtain a strip image pixel block set; Based on the set of pixel blocks in the strip image, the strip image segments under the same window number are rearranged and stitched according to the scan record number order and column number order to determine the strip pixel sequence in each angle window and obtain the remapped strip data.

7. The high-speed real-time line laser three-dimensional imaging method according to claim 6, characterized in that, Based on the readable pixel data, adjacent and consecutive readable pixels under the same window number in both column and row order are aggregated to obtain a striped image pixel block set, including: Based on the readable pixel data, the readable pixels under the same window number are compared in column order and row direction to identify pixel association groups that are continuous in both dimensions, thus obtaining pixel association data. Based on pixel association data, pixels that belong to the same column order direction but have discontinuities in row arrangement are linked across rows to fill in pixel gaps and obtain row completion data. By integrating the row-direction pixels in the row-direction completion data with the pixels in the adjacent column order, a pixel combination that is continuously distributed in both column order and row order is formed, resulting in continuous pixel combination data. Based on the continuous pixel combination data, the boundary range, column span, and row span of each pixel combination are summarized to determine the pixel block set of the local segment space of the strip, thus obtaining the pixel block set of the strip image.

8. The high-speed real-time line laser three-dimensional imaging method according to claim 7, characterized in that, Based on the angle window index and column order index in the remapped strip data, the three-dimensional coordinates are calculated and three-dimensional strip coordinates are formed to obtain real-time three-dimensional imaging data, including: Based on the remapped strip data, the angle window index and column order index are combined to calculate the joint index position of each pixel block in the angle domain and column order domain, thus obtaining the joint index data; Based on the combined index data, the angle index is converted into an angle displacement parameter, the column order index is converted into a column position parameter, and the two-dimensional position value of each pixel block is calculated to obtain the two-dimensional position data. Based on the joint index data, the angular displacement parameter is used as the height derivation factor, and the column position parameter is used as the spatial expansion factor to calculate the spatial height value of each pixel block and obtain the height derivation data. Based on the two-dimensional position data and height derivation data, the two-dimensional position value and spatial height value of each pixel block are integrated to determine the spatial coordinates of the strip pixels, thus obtaining the three-dimensional strip coordinates.

9. The high-speed real-time line laser three-dimensional imaging method according to claim 8, characterized in that, Based on the two-dimensional position data and height derivation data, the two-dimensional position values ​​and spatial height values ​​of each pixel block are integrated to determine the spatial coordinates of the strip pixels, thus obtaining the three-dimensional strip coordinates, including: Based on the two-dimensional position data, the column position parameters and row position parameters are classified to determine the planar position of each pixel block in the two-dimensional coordinate plane, thus obtaining the planar position data. Based on the height derivation data, the height derivation values ​​corresponding to each plane position in the plane position data are matched, and the plane positions are mapped to the height derivation values ​​to obtain height matching data; Based on the height matching data, the plane positions in the planar position are combined with their corresponding height derivation values ​​to form three-dimensional position units, thus obtaining position integration data; Based on the location integration data, the three-dimensional location units are arranged according to the order of pixel blocks in the remapped strip data to determine the three-dimensional location sequence of strip pixels and obtain the three-dimensional strip coordinates.

Images