Large-area array structure pattern segmentation and multi-view field splicing processing method

Abstract

The application discloses a large-area array structure pattern segmentation and multi-view field splicing processing method, and relates to the technical field of laser processing. The method is based on DXF pattern files to extract processing object center coordinates and equivalent processing dimensions, and constructs a processing object set; a regular grid is established by using an effective view field of a galvanometer, and initial view field division is carried out; cross-border processing object detection is realized based on the construction of a minimum circumscribed rectangle R-tree space index; target view field division results are generated by combining overall grid translation optimization and local boundary self-adaptive adjustment; and processing subfiles and view field position information are generated, so that multi-view field splicing processing and seam suppression of the galvanometer scanning system and the motion platform are realized.

Description

Technical Field

[0001] This invention relates to the field of laser processing technology, and more specifically to a method for processing large-area array structure graphics segmentation and multi-field splicing. Background Technology

[0002] With the development of ultra-precision laser processing technology, large-area array microstructures are widely used in fields such as vacuum adsorption, biomedicine, microfluidics, and optical devices. Due to the limited effective processing range of galvanometer scanning systems, when the area to be processed exceeds the field of view of a single galvanometer, it is usually necessary to combine a motion platform with a multi-field stitching method to complete the processing of the entire pattern.

[0003] Existing methods for large-area multi-field stitching typically use DXF graphic files as the data source and generate multiple processing fields of view by dividing the overall graphic into a grid according to fixed dimensions. However, since there is no correlation between the processing objects in the DXF graphic and the preset grid, it is easy for a single processing object to be segmented by multiple field boundaries during the field of view division process, resulting in cross-boundary distribution of processing objects. For cross-boundary processing objects, existing methods usually handle them by repeated processing, truncation processing, or manual adjustment of the divided regions. This not only increases the workload of processing preparation but also easily leads to repeated processing or omission of processing in local areas, affecting the integrity and consistency of the processed objects.

[0004] Furthermore, traditional field-of-view segmentation methods lack an automatic optimization mechanism for cross-domain processing objects. This leads to seam defects during the splicing process of multiple generated processing fields of view, reducing the processing quality of large-area array structures. Therefore, there is an urgent need for a graphic segmentation method that can optimize field-of-view segmentation by incorporating the spatial distribution characteristics of the processing objects and generate graphics suitable for multi-field splicing processing. Summary of the Invention

[0005] The purpose of this invention is to provide a method for large-area array structure graphic segmentation and multi-field stitching processing. By parsing DXF graphic files, the method extracts the center coordinates and equivalent processing dimensions of the processed objects and constructs a set of processed objects. This transforms the original processed graphics from graphic data into computable and processable parameterized data, providing a unified data foundation for subsequent field-of-view segmentation and boundary optimization. Furthermore, based on the minimum bounding rectangle of the processed objects, an R-tree spatial index is established to achieve rapid retrieval of the spatial relationship between the processed objects and the field-of-view boundary, and accurate identification of cross-boundary processed objects. This method reduces the processing time for large-scale processing while ensuring detection accuracy. The computational cost of object search is reduced; then, based on the cross-boundary detection results, the regular mesh is optimized by overall translation, and combined with local boundary adaptive adjustment, which effectively reduces the number of cross-boundary processing objects, avoids repeated processing or omission of processing objects caused by the segmentation of the field of view boundary, and improves the rationality of the field of view division results and the integrity of the processing objects; finally, based on the optimized field of view division results, a processing sub-file is generated and the galvanometer scanning system and motion platform are controlled to complete the multi-field stitching processing, which provides a unified and accurate processing basis for the subsequent implementation of the stitching seam suppression strategy, thereby improving the stability and processing consistency of multi-field stitching processing of large-area array structures.

[0006] To achieve the above objectives, the present invention provides the following technical solution: In a first aspect, the present invention provides a method for large-area array structure graphic segmentation and multi-view splicing processing, the method comprising: Obtain the DXF graphic file corresponding to the array structure to be processed, parse the DXF graphic file, remove non-machined entities, perform geometric recognition on the machined entities, and extract the center coordinates and equivalent machined dimensions of the machined object. Construct a set of machining objects based on the center coordinates and equivalent machining dimensions: ; Among them, (x i ,y i Let be the center coordinates of the i-th processing object. The equivalent machining dimension of the i-th machining object; A regular grid is established based on the effective processing field of view of the galvanometer, and the initial field of view is divided for the set of processing objects to obtain multiple initial fields of view. For each processing object, establish a minimum bounding rectangle: ; Among them, O i For any object to be processed, consider it as a circular object with its center at (x). i ,y i Equivalent machining radius r i =d proc / 2,dproc The equivalent machining dimension of the object being machined; An R-tree spatial index is established based on the minimum bounding rectangle, and cross-boundary detection is performed on each initial field of view boundary to determine the cross-boundary processing object; Based on the cross-boundary detection results, perform overall translation optimization on the regular mesh: ; Among them, Ω ij For the (i, j)th grid region, Δ x and Δ y These represent the translation amounts of the grid in the x and y directions, respectively. After overall translation optimization, adaptive adjustment is performed on the local boundaries of the processing objects that still cross boundaries to obtain the target field of view division results; Based on the target field of view division results, generate multiple corresponding processing sub-files and the field of view position information corresponding to each processing sub-file; Based on the processing sub-files and field-of-view position information, the galvanometer scanning system and motion platform are controlled to perform multi-field stitching processing, and a stitching seam suppression strategy is implemented during the stitching process to complete the processing of large-area array structures.

[0007] Secondly, the present invention also provides a large-area array structure pattern segmentation and multi-view splicing processing device, the device comprising: The file parsing module is used to obtain the DXF graphic file corresponding to the array structure to be processed, parse the DXF graphic file, remove non-processable entities, perform geometric recognition on the processed entities, and extract the center coordinates and equivalent processing dimensions of the processed object. The collection building module is used to construct a collection of machining objects based on the center coordinates and equivalent machining dimensions. ; Among them, (x i ,y i Let be the center coordinates of the i-th processing object. The equivalent machining dimension of the i-th machining object; The initial field-of-view module is used to establish a regular grid based on the effective processing field-of-view size of the galvanometer, and to divide the processing object set into initial fields of view to obtain multiple initial fields of view. The rectangle creation module is used to create a minimum bounding rectangle for each processing object. ; Among them, O i For any object to be processed, consider it as a circular object with its center at (x). i ,y i Equivalent machining radius r i =d proc / 2,d proc The equivalent machining dimension of the object being machined; An R-tree spatial index is established based on the minimum bounding rectangle, and cross-boundary detection is performed on each initial field of view boundary to determine the cross-boundary processing object; The translation optimization module is used to perform overall translation optimization on the regular mesh based on the cross-boundary detection results. ; Among them, Ω ij For the (i, j)th grid region, Δ x and Δ y These represent the translation amounts of the grid in the x and y directions, respectively. The boundary adjustment module is used to perform adaptive adjustment on the local boundaries of processing objects that still cross boundaries after the overall translation optimization, so as to obtain the target field of view division result; The information generation module is used to generate multiple corresponding processing sub-files and the field of view position information of each processing sub-file based on the target field of view division results. The splicing and processing module is used to control the galvanometer scanning system and motion platform to perform multi-field splicing and processing according to the processing sub-file and field of view position information, and to perform splicing seam suppression strategy during the splicing and processing process to complete the processing of large-area array structures.

[0008] Thirdly, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the large-area array structure pattern segmentation and multi-view splicing processing method provided in the first aspect.

[0009] Fourthly, the present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the large-area array structure pattern segmentation and multi-view splicing processing method provided in the first aspect.

[0010] Fifthly, the present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the large-area array structure graphic segmentation and multi-view splicing processing method provided in the first aspect.

[0011] The beneficial effects of this invention are as follows: By parsing DXF graphic files, this invention extracts the center coordinates and equivalent processing dimensions of the processed objects and constructs a set of processed objects, transforming the original processing graphics from graphic data into computable and processable parameterized data, providing a unified data foundation for subsequent field of view division and boundary optimization; furthermore, it establishes an R-tree spatial index based on the minimum bounding rectangle of the processed objects, enabling rapid retrieval of the spatial relationship between the processed objects and the field of view boundary and accurate identification of cross-boundary processed objects, reducing the computational load of searching large-scale processed objects while ensuring detection accuracy; subsequently, it performs overall translation optimization on the regular grid based on the cross-boundary detection results, combined with adaptive adjustment of local boundaries, effectively reducing the number of cross-boundary processed objects, avoiding repeated processing or omissions caused by the segmentation of processed objects by the field of view boundary, and improving the rationality of the field of view division results and the integrity of the processed objects; finally, it generates processing sub-files based on the optimized field of view division results and controls the galvanometer scanning system and motion platform to complete multi-field stitching processing, providing a unified and accurate processing foundation for the implementation of subsequent stitching suppression strategies, thereby improving the stability and consistency of multi-field stitching processing of large-area array structures.

[0012] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, the preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings. Attached Figure Description

[0013] Figure 1 This is a flowchart illustrating a method for large-area array structure graphic segmentation and multi-view splicing processing according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the seam morphology under different transition zone conditions according to an embodiment of the present invention; Figure 3 This is a schematic diagram showing the splicing effect and three-dimensional morphology of the seam under different processing strategies in an embodiment of the present invention; Figure 4 This is a schematic diagram illustrating the splicing effect under different layering methods in one embodiment of the present invention; Figure 5 This is a schematic diagram illustrating the splicing effect of different cyclic processing schemes in one embodiment of the present invention. Detailed Implementation

[0014] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0015] It should be noted that references to "an embodiment," "embodiment," "example embodiment," etc., in this specification refer to the described embodiment including specific features, structures, or characteristics; however, not every embodiment must include these specific features, structures, or characteristics. Furthermore, such expressions do not refer to the same embodiment. Moreover, when describing specific features, structures, or characteristics in conjunction with embodiments, whether or not explicitly described, it is indicated that incorporating such features, structures, or characteristics into other embodiments is within the knowledge of those skilled in the art.

[0016] Furthermore, the technical features involved in the different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0017] Example 1 This embodiment provides a method for large-area array structure graphic segmentation and multi-view splicing processing, such as... Figure 1 As shown.

[0018] S101: Obtain the DXF graphic file corresponding to the array structure to be processed, parse the DXF graphic file, remove non-processable entities, perform geometric recognition on the processed entities, and extract the center coordinates and equivalent processing dimensions of the processed object.

[0019] In this context, a large-area array structure refers to a two-dimensional fabricated pattern composed of multiple repeating microstructure units arranged according to a preset pattern. Its overall size typically exceeds the effective scanning range of a single galvanometer system, thus requiring it to be divided into multiple fields of view for separate fabrication. In this embodiment, the large-area array structure can be a micropillar array on the surface of a vacuum chuck, or it can be a micro-hole array, micro-pit array, micro-protrusion array, microlens array, or other two-dimensional microstructure arrays with regular arrangement characteristics. For example, when fabricating a uniformly distributed micropillar array on the surface of a 60mm × 60mm vacuum chuck, since the effective processing range of a single galvanometer system can only cover a portion of the area, it is necessary to first rationally divide the overall pattern and then complete the fabrication of each field of view separately.

[0020] To ensure that the processed graphics can be recognized by the computer and further divided into sections, this embodiment uses DXF (Drawing Exchange Format) as the data carrier for the graphics to be processed. DXF files can store different types of primitive information, such as circles, lines, polylines, splines, block references, and text annotations, and are a commonly used two-dimensional graphic description format in the laser processing field. Because DXF files generated by different CAD software differ in layer definitions, entity types, and coordinate precision, it is necessary to uniformly parse the DXF graphic files before formally dividing the graphics.

[0021] Specifically, the HEADER, TABLES, BLOCKS, and ENTITIES data segments in the DXF file are first read, and a graphic database is established. Then, all graphic elements in the ENTITIES data segment are traversed, and the entity type, layer, geometric dimensions, coordinate position, and attribute parameters corresponding to each graphic element are extracted. For hidden layers, auxiliary layers, annotation layers, and entities such as text, dimension annotations, center lines, and auxiliary lines used only for design descriptions, since they do not participate in the actual laser processing, they are uniformly identified as non-processing entities and directly deleted to avoid generating invalid data in the subsequent graphic division process and improve the efficiency of subsequent calculations.

[0022] Furthermore, to ensure that data generated by different CAD software has a unified data format, DXF files also need to be standardized. Standardization includes unifying the coordinate system, length units, layer attributes, and data precision. For example, if some DXF files use inches as the length unit while others use millimeters, all coordinates can be converted to millimeters using a preset conversion relationship. Similarly, if different software outputs have different decimal places, causing slight errors in the center coordinates of adjacent machined objects, the coordinate data can be uniformly retained according to a preset precision to ensure that subsequent geometric calculations are not affected by inconsistent data precision.

[0023] After deleting non-processing entities and standardizing the data, it is necessary to further identify the actual processing objects. Processing objects refer to the geometric entities that will ultimately be scanned by the laser beam to form the target microstructure. These can be either circles in DXF or standard processing units defined in the form of block references. When a processing object is represented as a circle, the center coordinates and radius can be directly read as its geometric parameters. When a processing object is represented as a block reference, the geometric center and dimension parameters in the block definition are further read and converted into a unified processing object description method, thus ensuring a consistent data interface for different types of primitives in subsequent processing flows.

[0024] After identifying all processing objects, it is necessary to further extract the center coordinates and equivalent processing dimensions of each object. The center coordinates uniquely determine the spatial position of the object within the entire processing area, serving as a crucial basis for subsequent mesh generation, spatial indexing, and boundary detection. The equivalent processing dimensions characterize the actual laser's effective range, which is related not only to the designed graphic dimensions but also requires consideration of factors such as the laser spot size, processing compensation, and the actual ablation range. Therefore, this embodiment does not directly use the designed dimensions but introduces processing compensation based on them to calculate the equivalent processing dimensions of the objects. This ensures that subsequent graphic segmentation better matches the actual processing boundaries, avoiding overlaps or omissions in the actual processing areas caused by segmentation based solely on the designed dimensions.

[0025] S102, construct a set of machining objects based on the center coordinates and equivalent machining dimensions.

[0026] The set of processing objects is used to uniformly describe the spatial distribution relationship among all objects to be processed within the entire processing area. Its mathematical expression is as follows: ; Among them, (x i ,y i Let be the center coordinates of the i-th processing object. Let be the equivalent machining dimension of the i-th machining object.

[0027] Through the above parameterized description method, the DXF file, which originally existed in graphical form, is converted into a unified data set, so that each processed object has unique position and size parameters, providing a unified data input for subsequent regular mesh division, spatial index establishment, and cross-boundary detection.

[0028] Furthermore, after constructing the set of processing objects, a sequential numbering system can be established based on the set. For example, the numbering can be completed sequentially according to the center coordinates of the processing objects, with priority given to the X direction and then to the Y direction. Alternatively, the processing objects can be numbered by partitioning the area where they are located, so as to facilitate quick location of specific processing objects and improve overall data management efficiency.

[0029] Through the above processing, the original DXF graphic is transformed from a two-dimensional design graphic into parametric machining data, enabling graphic data generated by different CAD software to be described using a unified data structure. At the same time, it removes the interference caused by non-machining entities to subsequent graphic segmentation, and fully considers the actual laser processing compensation factors, improving the accuracy of the spatial description of the machining object. This provides a reliable data foundation for subsequent regular mesh establishment, spatial index retrieval, and optimization of multi-view segmentation. It also avoids the problem of cumulative errors in subsequent field-of-view segmentation caused by inconsistent original graphic formats or inaccurate descriptions of machining object dimensions, thus improving the data consistency and subsequent processing efficiency of the entire graphic segmentation process.

[0030] S103: Establish a regular grid based on the effective processing field size of the galvanometer, divide the processing object set into initial fields of view, and obtain multiple initial fields of view.

[0031] The effective processing field of view of the galvanometer refers to the largest two-dimensional area that the galvanometer scanning system can continuously process while ensuring processing accuracy, focusing quality, and scanning stability. As the galvanometer deflection angle increases, the laser focal point position, spot size, and scanning distortion all change. Therefore, although the galvanometer can cover a larger scanning range, only the central area can guarantee high processing accuracy. In this embodiment, the size of the effective processing field of view is determined based on the calibration results of the processing equipment. For example, a single effective processing field of view can be determined as 10mm×10mm, 15mm×15mm, or 20mm×20mm, etc., based on the galvanometer focal length, field lens specifications, and processing accuracy requirements. When the size of the area to be processed exceeds a single effective processing field of view, the entire processing area needs to be divided into multiple fields of view that can be processed independently.

[0032] A regular mesh refers to a two-dimensional mesh structure formed by uniformly discretizing the entire processing area along the X and Y directions according to a preset mesh size. Each mesh corresponds to an initial processing field of view. The regular mesh not only serves as the basis for subsequent graphic segmentation but also as a reference coordinate for subsequent overall mesh optimization and local boundary adjustment. Therefore, the accuracy of establishing the regular mesh directly affects the overall quality of graphic segmentation.

[0033] Specifically, such as Figure 2 As shown, firstly, the maximum and minimum values ​​of the center coordinates of all processed objects in the processing object set are counted to determine the minimum envelope region corresponding to the entire processing area. Then, based on the effective processing field size of the galvanometer, the number of grids required in the X and Y directions are calculated respectively. Taking the lower left corner of the processing area as the initial grid origin, regular grids are established sequentially along the X and Y directions to finally form a two-dimensional grid matrix covering the entire processing area.

[0034] For example, when the processing area size is 60mm×60mm and the effective processing field of view size of the galvanometer is 10mm×10mm, a 6×6 regular grid can be established first, forming a total of 36 initial fields of view. If the processing area size cannot be divided by the effective processing field of view size of the galvanometer, the outermost grid size can be appropriately expanded according to the remaining size of the edge area, or the overall grid position can be adjusted according to the preset compensation strategy so that all processing objects can fall within the coverage area of ​​the regular grid.

[0035] To facilitate subsequent retrieval, each grid cell corresponds to a unique grid number, which can be represented using two-dimensional coordinates. For example, let the global bounding box of the micropillar set P be [X...]. min ,X max ]×[y min ,y max The width and height of the single-view grid are L respectively. x and L y The number of rows and columns of the regular quadrilateral grid can be determined as follows: ; The grid corresponding to the i-th row and j-th column can be represented as Ω. ij Thus, the boundary region of the (i,j)th grid cell can be constructed: ; Among them, Ω ij This represents the regular grid region located in the i-th row and j-th column, where i = 0, ..., n x -1,j=0,...,n y -1. Each regular grid corresponds to a unique processing field of view and has fixed boundary coordinates, center coordinates, and dimensional parameters. Using this method, the entire processing area can be represented as multiple adjacent but independent regular grids, establishing a unified data foundation for subsequent graphic partitioning.

[0036] Subsequently, the initial field of view needs to be divided based on the set of processing objects. Initial field of view division refers to determining the regular grid to which each processing object belongs based on its center coordinates and assigning the corresponding processing object to the appropriate field of view. In practice, the positional relationship between the center coordinates of the processing objects and the boundary coordinates of each regular grid can be compared. When the center coordinates of a processing object are within the range of a certain regular grid, the processing object is assigned to the corresponding regular grid; if multiple processing objects are located on the same regular grid, they collectively form the set of processing objects within that field of view.

[0037] It should be noted that only the initial field of view division is completed at this stage. This division is based on the center position of the processing object and does not determine whether the object is completely within the corresponding field of view. For example, if the center of a processing object is located within a regular grid, but its size is large enough that its edges extend beyond the current field of view boundary, then although its center coordinates still belong to the current field of view, the processing entity has actually crossed the boundary of an adjacent field of view. If processing is performed directly according to the initial field of view, it is easy to cause the processing object to be artificially segmented, resulting in duplicate processing or missed processing. Therefore, this embodiment uses the initial field of view division only as the initial result for subsequent cross-boundary detection, not the final processing field of view.

[0038] To further improve subsequent computational efficiency, after the initial field of view is defined, a field of view information table can be created. This table should include at least the field of view number, the coordinates of the field of view center, the coordinates of the field of view boundaries, the numbers of the processed objects included, and the quantity of processed objects. For example, for the field of view numbered Ω23, its information table can record the coordinates of the lower left corner, the upper right corner, the center position, and a list of all processed object numbers. This allows for quick acquisition of the data corresponding to the current field of view without having to traverse all processed objects again, improving the efficiency of subsequent spatial retrieval.

[0039] Furthermore, since a unified coordinate system is used during the establishment of the regular mesh, subsequent overall mesh translation optimization only requires updating the mesh boundary coordinates synchronously. There is no need to re-parse the DXF graphics or rebuild the processing object set, significantly reducing the computational load for subsequent optimization. Simultaneously, the unified regular mesh structure provides a standard reference boundary for local boundary adjustments, enabling overall and local optimizations to be completed collaboratively within the same coordinate system.

[0040] Through the above steps, the original discretely distributed processing objects are further organized into a regular grid structure with spatial topological relationships, and the initial field of view is divided. This transforms the entire processing area from a single graphic data into a field of view data structure that can be used for spatial optimization calculations. This not only ensures that the subsequent spatial indexing has a unified data foundation, but also provides accurate initial field of view information for cross-boundary processing object detection, overall grid translation optimization, and local boundary adaptive adjustment. It avoids the impact of chaotic initial field of view division on the overall graphic segmentation accuracy during subsequent multi-field stitching, and improves the data organization efficiency and computational stability of the entire large-area array structure graphic segmentation process.

[0041] S104, establish the minimum bounding rectangle for each processing object.

[0042] After the initial field of view is defined, it is necessary to further determine whether each processing object is completely located within its respective field of view to avoid the processing objects being segmented by the field of view boundary, which would affect the subsequent stitching processing quality. Therefore, this embodiment further establishes a spatial index structure for the processing objects and uses the spatial index to quickly complete the detection of cross-boundary processing objects.

[0043] The Minimum Bounding Rectangle (MBR) is an axis-aligned rectangle that completely encloses the target object and has the smallest area. It provides a unified description of the object's spatial extent using four boundary coordinates. Since subsequent cross-boundary detection focuses more on the object's spatial occupancy than its specific geometric contours, using the MBR significantly reduces the computational complexity of subsequent spatial calculations while maintaining spatial accuracy. For example, for a circular micro-cylinder object, the MBR can be directly calculated based on the object's center coordinates and equivalent processing dimensions. For objects represented using block references, the MBR can be established based on the maximum bounding area of ​​the corresponding geometric primitives, thus enabling different types of objects to be described using a unified data structure.

[0044] Specifically, in this embodiment, a corresponding minimum bounding rectangle is established for each processing object in the processing object set, and the establishment method is as follows: ; Among them, O i For any object to be processed, consider it as a circular object with its center at (x). i ,y i Equivalent machining radius r i =d proc / 2,d proc The equivalent machining dimension of the object being machined.

[0045] The above formula is used to determine the boundary range of the processing object in the X and Y directions. The established minimum bounding rectangle includes at least the coordinates of the left boundary, right boundary, lower boundary, and upper boundary. In this way, each processing object is further transformed from its original geometric primitive into a spatially enclosed region, providing a unified data foundation for subsequent spatial retrieval.

[0046] S105. Based on the minimum bounding rectangle, establish an R-tree spatial index, perform cross-boundary detection on each initial field of view boundary, and determine the cross-boundary processing object.

[0047] After establishing the minimum bounding rectangle for all processed objects, an R-tree spatial index is further constructed. An R-tree is a hierarchical index structure for spatial data organization. It can build a tree-like index based on the spatial inclusion relationships between minimum bounding rectangles, enabling fast retrieval of two-dimensional spatial objects. Compared to traversing all processed objects one by one for boundary checks, an R-tree can first perform a coarse screening using the minimum bounding rectangle of each node, then progressively narrow the search range, further judging only processed objects that may have spatial intersections. Therefore, it is particularly suitable for managing large-scale two-dimensional spatial objects.

[0048] In the specific implementation, the smallest bounding rectangle corresponding to each processing object is first used as the leaf node, and multiple processing objects with similar spatial positions are merged to form a parent node. Then, higher-level nodes are continuously built upwards until a unique root node is finally formed. Each parent node records the overall bounded area of ​​the smallest bounding rectangle corresponding to all its child nodes. Therefore, the entire R-tree actually forms a coarse-to-fine spatial organization structure. When it is necessary to query whether there is a processing object in a certain area, only the nodes that have spatial intersection with the query area need to be accessed, without traversing all processing objects, thus significantly improving the efficiency of spatial query.

[0049] After establishing the R-tree spatial index, cross-boundary detection needs to be performed based on the initial field of view boundaries. Cross-boundary detection refers to determining whether a processing object simultaneously occupies two or more adjacent fields of view, thereby identifying whether the processing object is a cross-boundary processing object. Cross-boundary processing objects are the main targets for subsequent overall mesh optimization and local boundary adjustment, and are also an important factor affecting the formation of seams. Therefore, the cross-boundary detection results directly determine the subsequent graphics optimization effect.

[0050] To avoid redundant calculations for all processed objects within the entire field of view, this embodiment first establishes boundary query regions based on the boundaries of each field of view. The boundary query region is a strip-shaped area extending outwards from the field of view boundary, and its width is determined by the maximum equivalent processing size of the processed object. Since only processed objects near the field of view boundary are likely to cross the boundary, establishing boundary query regions can directly exclude a large number of processed objects located within the field of view from subsequent detection, thereby further reducing the spatial search range and improving detection efficiency.

[0051] Subsequently, the boundary query region is input into the R-tree spatial index as a spatial query window. The R-tree first determines whether a node intersects with the query window based on the minimum bounding rectangle corresponding to each node at each level. If no spatial intersection exists, the node is pruned directly, and its child nodes are not visited again. If a spatial intersection exists, the search continues to the next level until all candidate processing objects are obtained. Because the R-tree uses a layer-by-layer filtering method, even if the processing region contains tens of thousands or even hundreds of thousands of processing objects, only a very small number of nodes need to be visited to complete the extraction of candidate objects, effectively reducing the complexity of spatial retrieval.

[0052] After obtaining candidate processing objects, a geometric determination is required to accurately identify cross-boundary processing objects. Geometric determination refers to judging whether a processing object is completely within the current field of view based on the positional relationship between its minimum bounding rectangle and the field of view boundary. If the entire minimum bounding rectangle of a processing object is within the field of view, the object is considered to belong to the current field of view. If any boundary of the minimum bounding rectangle extends beyond the current field of view boundary, the object is determined to be a cross-boundary processing object, and its corresponding field of view number, the direction of boundary crossing, and adjacent field of view numbers are recorded, providing data for subsequent overall mesh optimization.

[0053] For example, when the center of a micro-pillar being processed is located near the right side of the current field of view, and the right boundary of its minimum bounding rectangle has exceeded the current field of view boundary, it means that the processing object occupies both the current field of view and the adjacent field of view to the right. The system can then automatically mark it as a horizontal cross-boundary processing object. Similarly, when the minimum bounding rectangle of a processing object exceeds both the X and Y direction boundaries, it can be determined that it belongs to a corner cross-boundary processing object, and the corresponding cross-boundary type can be recorded.

[0054] To facilitate subsequent overall mesh optimization and local boundary adjustments, this embodiment further establishes a cross-boundary processing object information table. The information table includes at least the processing object number, its initial field of view number, the type of boundary it crosses, the adjacent field of view numbers, and the coordinates of its smallest bounding rectangle. During subsequent overall mesh optimization, it is unnecessary to traverse all processing objects again; optimization calculations only need to be performed on the cross-boundary processing object information table, significantly improving overall optimization efficiency.

[0055] Through the above processing, a unified expression of the spatial range of the processing object is achieved using the minimum bounding rectangle. Furthermore, the R-tree spatial index is combined to complete the rapid screening of candidate processing objects. All cross-boundary processing objects are accurately identified through geometric judgment. This transforms the traditional boundary detection process, which requires traversing all processing objects, into a fast query process based on spatial index. This not only reduces the spatial search computation in the segmentation of large-scale array structure graphics, but also accurately obtains the cross-boundary information required for subsequent overall mesh translation optimization and local boundary adaptive adjustment, improving the efficiency and accuracy of cross-boundary detection and providing reliable data support for subsequent optimization of target field of view division results.

[0056] S106, Based on the cross-boundary detection results, perform overall translation optimization on the regular grid.

[0057] After completing the cross-boundary processing object detection, all cross-boundary processing object information that needs optimization can be obtained. However, what is obtained at this time is only the division result formed based on the initial regular mesh. Since the regular mesh is established by uniformly dividing the effective processing field of view of the galvanometer without considering the actual spatial distribution of the processing objects, there may still be a large number of processing objects being segmented by the field of view boundary. If the processing file is generated directly based on this initial field of view division result, multiple processing objects will need to be processed across adjacent fields of view, which not only increases the number of stitching areas, but also easily causes problems such as repeated processing, missed processing, and discontinuous edge contours at the stitching boundaries. Therefore, this embodiment further performs overall translation optimization on the regular mesh, and combines local boundary adaptive adjustment on the basis of overall optimization, so that the field of view boundary avoids the processing objects as much as possible, thereby achieving further optimization of the graphic segmentation result.

[0058] The overall translation optimization refers to maintaining the regular grid size and the topological relationship between adjacent grids, while only changing the initial position of all grids in the processing area. This redistributes processing objects that were originally located near the field of view boundary into the field of view, thereby reducing the number of cross-boundary processing objects. Since the relative positions between the grids remain unchanged during the overall translation process, it does not change the effective processing range of the galvanometer, nor does it add new processing areas; it only changes the correspondence between the field of view boundary and the processing objects.

[0059] Specifically, after the initial field of view is divided, the entire regular mesh is used as the optimization object, and different translation amounts are applied along the X and Y directions respectively. The translated regular mesh can be updated according to the following formula: ; Among them, Ω ij For the (i, j)th grid region, Δ x and Δ yThese represent the translation amounts of the grid in the x and y directions, respectively. During the overall translation, the dimensions of each grid remain unchanged, and adjacent grids maintain continuous connectivity; only the positions of all grids change synchronously.

[0060] It should be noted that the overall translation amount is not arbitrarily selected, but rather optimized through a preset search range. In this embodiment, search intervals in the X and Y directions can be established based on the effective processing field of view of the galvanometer. For example, Δx and Δy can be changed according to a preset step size to re-perform cross-boundary detection at different translation positions, and the number of cross-boundary processing objects can be counted. After all search positions have been calculated, the overall translation amount with the fewest cross-boundary processing objects is determined as the target overall translation amount. To further balance computational efficiency, the search step size can be adaptively adjusted according to the distribution density of processing objects. For example, a smaller search step size can be used in areas with dense processing objects, while a larger search step size can be used in areas with more uniform processing object distribution, thereby balancing optimization accuracy and computational efficiency.

[0061] For example, when the initial rule grid detects 132 cross-boundary processing objects, the system can perform overall translation in the X and Y directions according to preset step sizes. After each overall translation, a cross-boundary detection is re-executed, and the number of cross-boundary processing objects is re-counted. When the overall translation reaches a certain position and the number of cross-boundary processing objects decreases to 89, the current translation amount is recorded. After continuing the entire search, if no further reduction in the number of cross-boundary objects is found, the translation position is taken as the overall optimization result.

[0062] To ensure that the coordinates of each field of view still correctly correspond to the processing area after the overall translation, it is necessary to simultaneously update the coordinates of all field of view boundaries and the coordinates of the field of view center. Since the position of the processing object itself remains unchanged, only the field of view coordinate information is updated during the overall translation process. There is no need to re-parse the DXF graphics file or rebuild the processing object set. The new cross-boundary detection can be completed simply by re-performing a spatial query based on the new grid position, which greatly reduces the amount of data processing in the overall optimization process.

[0063] S107, after overall translation optimization, adaptive adjustment is performed on the local boundaries of the processing objects that still cross boundaries to obtain the target field of view division results.

[0064] After overall translation optimization, most cross-boundary processing objects have been readjusted to within a single field of view. However, since the spatial distribution of processing objects is usually somewhat random, some processing objects may still be located near the field of view boundary, so there are still a few cross-boundary processing objects. If the overall translation method is continued for optimization, it may cause new cross-boundary objects to reappear in other areas. Therefore, this embodiment further adopts local boundary adaptive adjustment to perform fine optimization on the remaining cross-boundary processing objects.

[0065] Local boundary adaptive adjustment refers to adjusting the position of the local field of view boundary where there are still cross-boundary processing objects, without changing the positional relationship of regular grids in other areas. This further reduces the number of cross-boundary processing objects while maintaining the stability of the overall regular grid structure. Since local adjustment only applies to the common boundary between two adjacent fields of view, the adjustment range is small, and the impact on the overall field of view layout is limited.

[0066] Specifically, firstly, the common boundary of existing cross-border processing objects is determined based on the cross-border processing object information table; then, using this common boundary as the adjustment object, the boundary position is moved to both sides within the allowable adjustment range. During the boundary adjustment process, the size of one field of view increases accordingly, while the size of its adjacent field of view decreases synchronously, and the adjustment amounts of the two remain consistent, so that the overall coverage of adjacent fields of view remains unchanged, avoiding changes in the overall processing area.

[0067] Furthermore, to ensure that the adjusted field of view still meets the effective processing range requirements of the galvanometer, this embodiment sets constraints on the adjustment range of the local boundary. This ensures that the local boundary can avoid cross-boundary processing objects without causing any field of view size to exceed the allowable processing range of the galvanometer. For example, if moving the boundary a certain distance to the left allows the original cross-boundary processing objects to enter the left field of view, and the size of the left field of view still meets the processing requirements, then the current adjustment result is retained; if continuing the adjustment would cause the left field of view size to exceed the allowable range, then the adjustment is terminated, and another adjustment direction is selected or the current boundary position is maintained.

[0068] During local boundary adjustment, it is also necessary to synchronously update the boundary coordinates, center coordinates, and adjacency information of two adjacent fields of view, so that subsequent processing file generation can still be completed based on the latest field of view data. To avoid repeated adjustments, this embodiment can also establish a local optimization record table to record the boundary number and adjustment amount that have been adjusted. For boundaries that have met the optimization requirements, they will not participate in subsequent adjustment calculations again, thereby further improving the overall optimization efficiency.

[0069] For example, when a processing object crosses the common boundary between fields of view Ω23 and Ω24, the system first determines the adjustment distance required for the processing object to enter the left or right field of view. If moving the common boundary to the right by 0.18mm will allow the processing object to be completely located in the left field of view, while the right field of view still meets the effective processing range of the galvanometer, then the system will automatically adjust the common boundary to the right by 0.18mm and simultaneously update the boundary coordinates, center coordinates, and field of view number information corresponding to Ω23 and Ω24, so that the processing object changes from a cross-boundary processing object to a single field of view processing object.

[0070] After all local boundary adjustments are completed, the system re-executes cross-boundary detection based on the updated field of view boundaries. If new cross-boundary processing objects still exist, local optimization continues. When the number of cross-boundary processing objects meets the preset optimization requirements or further adjustments cannot reduce the number of cross-boundary objects, optimization stops, and the current field of view division result is determined as the target field of view division result.

[0071] Through the above processing, the overall mesh translation optimization is first used to fully leverage the overall adjustment capability of the regular mesh, reducing most cross-boundary processing objects without changing the mesh topology. Then, for the cross-boundary processing objects that remain after the overall optimization, local boundary adaptive adjustment is further performed to realize a two-layer optimization mechanism that combines overall optimization and local optimization. This makes the final field of view boundary more closely match the actual spatial distribution of the processing objects, effectively reducing the situation where processing objects are divided by the field of view boundary, improving the rationality and completeness of the target field of view division results, providing a reliable data foundation for the subsequent generation of processing sub-files and the implementation of high-quality multi-field stitching processing, while reducing the number of stitching regions and improving the overall processing efficiency and graphic processing consistency.

[0072] S108, Based on the target field of view division results, generate multiple corresponding processing sub-files and the field of view position information corresponding to each processing sub-file.

[0073] After obtaining the target field-of-view segmentation results, it is necessary to further convert the optimized graphic segmentation results into data files that the processing equipment can directly recognize, so as to control the galvanometer scanning system and motion platform to complete the actual processing. Therefore, this embodiment generates multiple processing sub-files based on the target field-of-view segmentation results and establishes the field-of-view position information corresponding to each processing sub-file, realizing the conversion from graphic segmentation results to processing execution data.

[0074] Among them, a processing sub-file refers to an independent processing file formed by reorganizing all processing objects belonging to the same target field of view according to the target field of view division results. Each processing sub-file contains only the graphic data to be processed within the corresponding target field of view. It can be directly loaded by the laser processing software and control the galvanometer system to complete the scanning processing of the corresponding field of view. The field of view position information is used to describe the actual position of the processing area corresponding to each processing sub-file in the entire processing coordinate system. It includes at least the field of view number, the coordinates of the field of view center, the coordinates of the field of view boundary, the positioning coordinates of the motion platform, and the numbers of adjacent fields of view, to ensure that the motion platform can accurately move to the corresponding processing area.

[0075] Specifically, the system first traverses the target field of view division results, establishes an independent data buffer for each target field of view, and writes the corresponding processing objects into the corresponding buffer one by one according to the relationship of the processing objects; then, it regenerates the corresponding DXF processing file or other processing format files supported by the device according to the processing objects in the buffer, so that each processing sub-file contains only the processing objects corresponding to its target field of view, and there is no situation of duplicate records across fields of view.

[0076] Furthermore, while generating the machining sub-file, a field of view information index table is also established simultaneously. This index table establishes a one-to-one correspondence between the machining sub-file and the motion platform's positioning information. For example, field of view number Ω23 corresponds to machining sub-file File23, and also to the motion platform coordinates (X23, Y23) and the galvanometer scanning center coordinates C23. When the machining equipment reads File23, it can automatically drive the motion platform to the corresponding position based on the field of view information index table, using the galvanometer scanning center coordinates as the current galvanometer machining origin, thus ensuring that the machining graphic accurately corresponds to the actual position of the workpiece.

[0077] S109 controls the galvanometer scanning system and motion platform to perform multi-field stitching processing based on the processing sub-file and field of view position information, and performs a stitching seam suppression strategy during the stitching process to complete the processing of large-area array structures.

[0078] After the processing sub-files are generated, the multi-field stitching processing stage can begin. Multi-field stitching processing refers to the process where the motion platform moves sequentially to each target field of view along a predetermined path. After each target field of view is reached, the galvanometer scanning system scans and processes the graphic corresponding to the current processing sub-file. Once the current field of view is processed, the platform moves to the next target field of view, and so on, until all target fields of view are processed, thus achieving continuous processing of the entire large-area array structure.

[0079] To further reduce seams at adjacent field-of-view stitching locations, this embodiment implements a seam suppression strategy during processing. The seam suppression strategy refers to comprehensively optimizing multiple aspects, such as the processing path, processing area, and processing sequence, to address locations prone to seams during multi-field-of-view processing, resulting in a more continuous and uniform final processing outcome.

[0080] Specifically, this embodiment first establishes a transition zone. The transition zone refers to a region of a certain width located near the common boundary of two adjacent target fields of view. Its main function is to create a processing transition between adjacent fields of view, reducing contour discontinuities caused by energy abrupt changes or positioning errors at the boundary. Based on the preset transition zone width, corresponding transition zones can be extended into the interior of the two adjacent target fields of view. During processing, priority is given to ensuring the continuity of the processing trajectory within the transition zone with the processing trajectory of the adjacent fields of view, thereby reducing processing abrupt changes at the splicing boundary and improving overall processing continuity.

[0081] Furthermore, to avoid continuous heat accumulation in local areas due to continuous processing of adjacent fields of view, this embodiment plans the multi-field processing sequence based on the adjacency relationship between target fields of view. The so-called adjacency relationship refers to whether two target fields of view have a common boundary. When two target fields of view have a common boundary, they are considered to be adjacent fields of view; if they do not have a common boundary, they are considered to be non-adjacent.

[0082] In practice, a field-of-view adjacency matrix can be established based on the target field-of-view boundary, and the entire target field of view can be divided into at least two non-adjacent processing fields of view according to the adjacency matrix. For example, a checkerboard grouping method can be used to divide all target fields of view into a first processing group and a second processing group, so that no two target fields of view within the same processing group have a common boundary. Then, the processing of all target fields of view in the first processing group is completed first, and then the processing of the second processing group is completed, allowing sufficient cooling time between adjacent target fields of view, thereby reducing the heat accumulation phenomenon caused by local continuous processing.

[0083] Furthermore, for larger array structures, a serpentine or reciprocating scanning path can be used to plan the motion platform's movement route. This allows the motion platform to move to the next target field of view with the shortest possible distance after completing processing of the current target field of view, reducing idle travel time and improving overall processing efficiency. Simultaneously, to further reduce positional errors caused by frequent starts and stops of the motion platform, an optimal movement path can be established based on the spatial position of each target field of view. This ensures that the processing sequence meets adjacency constraints while minimizing the overall movement path length. In addition to optimizing the processing sequence, this embodiment further employs a layered processing method to reduce the impact of seams. Layered processing refers to dividing the target processing depth into multiple processing layers, with each layer completing the processing of the entire target field of view separately, rather than processing the entire depth of a single target field of view at once. For example, when the target processing depth is 60μm, it can be divided into six layers, each processing 10μm; after the first layer completes processing of the entire target field of view, the process returns to the first target field of view to begin processing the second layer, and so on, completing all processing layers sequentially. Because each layer has a shallow processing depth, it effectively reduces local heat input and improves processing consistency.

[0084] To further discretize the splicing boundary positions, this embodiment also employs a dynamic mesh offset strategy. Dynamic mesh offset refers to changing the initial position of the regular mesh between different processing cycles, so that the splicing boundaries corresponding to adjacent processing cycles no longer coincide, thereby dispersing the splicing error to different positions and avoiding the continuous accumulation of errors during multi-layer processing.

[0085] Specifically, after completing the current processing cycle, the initial mesh origin corresponding to the next processing cycle is updated, and the cycle offsets corresponding to the X and Y directions are calculated respectively, so that the next round of processing forms a new target field of view division result. Since the splicing boundary positions corresponding to different processing layers change, the final splicing error will be evenly distributed among multiple processing layers, effectively reducing the probability of obvious splicing at a single position and improving the uniformity and surface quality of the overall processed surface.

[0086] For example, in the first processing cycle, the splicing boundary is located at the center of the processing area; in the second processing cycle, the splicing boundary is shifted a certain distance in the X direction by dynamic offset; in the third processing cycle, it is further shifted in the Y direction, so that different processing layers form staggered splicing boundaries, thereby avoiding the continuous formation of processing seams in the same position.

[0087] After completing all processing cycles, the system determines whether all processing is complete based on the processing status information corresponding to all processing sub-files. When all target fields of view and all processing layers are processed, the multi-field stitching processing of the entire large-area array structure is completed, and the final processing result is obtained.

[0088] Through the above implementation methods, the target field of view division results are further converted into processing sub-files that can be directly executed by the processing equipment. Combined with the field of view position information, collaborative control between the motion platform and the galvanometer scanning system is achieved. Simultaneously, through the synergistic effect of multiple seam suppression strategies such as transition zone optimization, processing sequence optimization, layered processing, and dynamic mesh offset, the stitching boundaries between adjacent target fields of view are effectively optimized. This not only improves the continuity and consistency of multi-field stitching processing of large-area array structures but also reduces the impact of heat accumulation, positioning errors, and repeated seam overlap on processing quality, achieving integrated control from graphic optimization to processing execution. Thus, a complete data processing and processing execution closed loop is formed from original DXF graphic acquisition, processing object parameterization, regular mesh division, cross-boundary detection, target field of view optimization, processing sub-file generation, to multi-field stitching processing, giving the entire processing flow good continuity, stability, and engineering application value.

[0089] Example 2 To verify the effectiveness of the aforementioned large-area array structure pattern segmentation and multi-field stitching processing method, a verification experiment was conducted using a pre-built multi-field laser processing platform. The experimental platform consisted of a picosecond laser processing system, a galvanometer scanning system, and a two-dimensional motion platform. First, the pattern segmentation method described in the above embodiment was used to segment the array structure to be processed, generating multiple processing sub-files and corresponding field-of-view center coordinate sequences. Then, the motion platform completed the positioning of each processing field of view based on the field-of-view center coordinates. The galvanometer scanning system loaded the corresponding processing sub-files to complete single-field scanning processing. After all fields of view processing were completed, the processing results were morphologically measured and performance evaluated to verify the effectiveness of the proposed pattern segmentation method and multi-field stitching processing strategy.

[0090] To ensure comparability among the experiments, all experiments used the same single-view processing parameters, varying only the graphic segmentation area, transition zone width, multi-view processing sequence, layered processing method, and dynamic control strategy for the splicing boundary. Specifically, the PI tape mask preparation parameters were: average power Pavg = 3W, scan overlap rate SOR = 90%, line overlap rate LOR = 60%, defocusing amount DF = 0μm, and processing times Np = 7; the micropillar array processing parameters were: average power Pavg = 12W, scan overlap rate SOR = 90%, line overlap rate LOR = 80%, and defocusing amount DF = +100μm; the in-situ polishing parameters were: average power Pavg = 2.1W, defocusing amount DF = 0μm, and processing times Np = 3. Except for the experimental variables, all other processing conditions remained consistent to ensure that the experimental results only reflected the impact of the splicing boundary control strategy on the processing quality. During the experiment, the main evaluation indicators included the total travel distance L of the platform, the minimum processing time interval Δtmin between adjacent fields of view, the maximum width Wmax of the seam, the maximum depth Dmax of the seam, and the height H of the micropillar. The method of this application was comprehensively evaluated from three aspects: processing efficiency, seam quality, and array forming consistency.

[0091] First, the effect of transition zone width on seam suppression was verified. The experiment used a 2×2 field-of-view splicing method, with each field of view having a processing area of ​​10mm×10mm. The processing was performed four times in each field of view. Multiple sets of experiments were conducted with different transition zone widths (Wt), and the seam morphology obtained from each set of experiments was measured. The experimental results are as follows: Figure 2 As shown, Figure 2This diagram illustrates the seam morphology under different transition zone conditions. Experimental results show that when no transition zone is set (Wt=0μm), there is significant repeated processing at the boundaries of adjacent fields of view, resulting in severe heat accumulation in local areas. The maximum seam width, Wmax, was measured to be approximately 38.55μm. As the width of the transition zone gradually increases, the repeated heating effect is effectively alleviated, and the seam width decreases significantly. When Wt=25μm, the maximum seam width drops to approximately 23.05μm. Further increasing the transition zone width affects boundary continuity due to insufficient coverage of adjacent fields of view. When Wt reaches 80μm, the maximum seam width increases to approximately 78.66μm, and a distinct stepped boundary appears. These experimental results demonstrate that a properly set transition zone can effectively reduce heat accumulation caused by repeated processing of adjacent fields of view, effectively reducing the seam width while maintaining boundary continuity. Therefore, subsequent experiments used Wt=25μm as the transition zone width.

[0092] Subsequently, the processing sequence for multiple fields of view was verified. The experiment used a 3×3 field-of-view stitching method, with each field of view having a processing area of ​​5mm×5mm. The processing was performed four times in each field of view, using both a serpentine and symmetrical processing sequence. The total platform travel distance L, the minimum processing time interval Δtmin between adjacent fields of view, and the maximum seam width Wmax were statistically analyzed. The experimental results are as follows: Figure 3 As shown, Figure 3 This diagram illustrates the splicing effect and 3D morphology of the seams under different processing strategies. Experimental results show that the total platform travel distance under the serpentine processing sequence is 45mm, less than the 69.46mm corresponding to the symmetrical processing sequence, indicating that the serpentine processing sequence has higher platform movement efficiency. However, the minimum processing interval between adjacent fields of view is only 22s, while the symmetrical processing sequence reaches 107s, allowing for more sufficient heat dissipation time in the seam area. Therefore, the maximum seam width is reduced from 24.68μm to 19.36μm, a reduction of approximately 21.6%. These experimental results demonstrate that although the symmetrical processing sequence increases the platform travel distance, it effectively reduces heat accumulation caused by continuous processing of adjacent fields of view, improving the processing quality of the seam area. Therefore, subsequent experiments all used the symmetrical processing sequence to complete multi-field splicing processing.

[0093] Furthermore, the layered processing and seam discretization strategy were validated. A 60mm × 60mm micropillar array was used as the processing object. This processing area exceeds the effective processing range of a single galvanometer. During the experiment, the processing parameters for each layer were kept consistent, only the field-of-view division area corresponding to each processing layer was changed. Two layering methods were used: Method 1 used a 30mm × 30mm field-of-view division for all three processing layers; Method 2 used three different field-of-view division methods: 30mm × 30mm, 20mm × 20mm, and 12mm × 12mm, causing spatial misalignment of the splicing boundaries of different processing layers. The experimental results are as follows: Figure 4 As shown, Figure 4 The diagram illustrates the splicing effects under different layering methods. In method one, because the seam positions of the three processing layers are completely identical, local repetitive processing accumulates continuously, ultimately resulting in a maximum seam width (Wmax) of 30.72 μm and a maximum depth (Dmax) of 11.77 μm. In method two, the seam positions of each processing layer become discretely distributed, ultimately reducing Wmax to 18.48 μm and Dmax to 5.75 μm, representing reductions of approximately 39.8% and 51.1%, respectively. The experimental results demonstrate that by changing the field-of-view division scale corresponding to different processing layers, causing the seam positions to be discretely distributed between layers, the continuous accumulation of seam errors during multi-layer processing can be effectively reduced, improving the processing quality of the spliced ​​area.

[0094] Finally, the dynamic control strategy for the splicing boundary was verified. During the experiment, the area layering scheme within a single processing cycle remained unchanged, with different offsets applied only to the initial mesh origin between adjacent processing cycles. In Scheme 1, the same initial mesh position was used for each processing cycle, i.e., the offset was (0, 0); in Scheme 2, (0, 0), (1, 1), and (2, 2) mm were used as the mesh offsets for each processing cycle, respectively, causing the splicing seam positions corresponding to different processing cycles to migrate as a whole. The experimental results are as follows: Figure 5 As shown, Figure 5 The diagram illustrates the splicing effects of different cyclic processing schemes. In Scheme 1, since the splicing positions of each processing cycle completely overlap, each splice is processed three times, and the maximum splice width Wmax is measured to be 39.47μm and the maximum depth Dmax is 13.01μm. In Scheme 2, by dynamically offsetting between cycles, the splicing positions of each cycle are staggered, and Wmax is reduced to 21.27μm and Dmax is reduced to 6.55μm, representing reductions of approximately 46.1% and 49.7%, respectively. Furthermore, 20 micropillars were randomly selected from the two sets of experimental micropillar arrays for statistical analysis of their height and top diameter. The results showed that the average heights of the two sets of micropillars were 22.59 μm and 23.63 μm, respectively, and the average top diameters were 350.29 μm and 350.54 μm, respectively. Both results showed small errors compared with the predicted heights, and the top contours were continuous and complete without any obvious breaks or misalignments. This indicates that the dynamic mesh offset mainly affects the spatial position control of the seams and does not damage the overall forming quality and dimensional consistency of the micropillar array.

[0095] The experimental results show that by employing graphic preprocessing, parameterized representation of the processing object, regular mesh generation, R-tree spatial index retrieval, overall mesh translation optimization, and local boundary adaptive adjustment to complete the target field of view division, and combining it with seam suppression strategies such as transition zone compensation, symmetrical processing sequence, layered processing, and dynamic mesh offset, the seam width and depth generated during multi-field stitching processing can be effectively reduced, while maintaining the consistency of the overall forming size of the micro-pillar array. This verifies that the method of this application is applicable to the graphic segmentation and multi-field stitching processing of large-area array structures that exceed the effective processing range of a single galvanometer, and can meet the application requirements of high-quality laser processing of large-area microstructures.

[0096] Based on the same inventive concept, this application also provides a device for large-area array structure pattern segmentation and multi-field stitching processing to implement the aforementioned large-area array structure pattern segmentation and multi-field stitching processing method. The solution provided by this device is similar to the solution described in the above method. Therefore, the specific limitations of one or more embodiments of the large-area array structure pattern segmentation and multi-field stitching processing device provided below can be found in the limitations of the large-area array structure pattern segmentation and multi-field stitching processing method described above, and will not be repeated here.

[0097] In one embodiment, a large-area array structure pattern segmentation and multi-view splicing processing device is provided, the device comprising: The file parsing module is used to obtain the DXF graphic file corresponding to the array structure to be processed, parse the DXF graphic file, remove non-processable entities, perform geometric recognition on the processed entities, and extract the center coordinates and equivalent processing dimensions of the processed object. The collection building module is used to construct a collection of machining objects based on the center coordinates and equivalent machining dimensions. ; Among them, (x i ,y i Let be the center coordinates of the i-th processing object. The equivalent machining dimension of the i-th machining object; The initial field-of-view module is used to establish a regular grid based on the effective processing field-of-view size of the galvanometer, and to divide the processing object set into initial fields of view to obtain multiple initial fields of view. The rectangle creation module is used to create a minimum bounding rectangle for each processing object. ; Among them, O i For any object to be processed, consider it as a circular object with its center at (x). i ,y i Equivalent machining radius ri =d proc / 2,d proc The equivalent machining dimension of the object being machined; An R-tree spatial index is established based on the minimum bounding rectangle, and cross-boundary detection is performed on each initial field of view boundary to determine the cross-boundary processing object; The translation optimization module is used to perform overall translation optimization on the regular mesh based on the cross-boundary detection results. ; Among them, Ω ij For the (i, j)th grid region, Δ x and Δ y These represent the translation amounts of the grid in the x and y directions, respectively. The boundary adjustment module is used to perform adaptive adjustment on the local boundaries of processing objects that still cross boundaries after the overall translation optimization, so as to obtain the target field of view division result; The information generation module is used to generate multiple corresponding processing sub-files and the field of view position information of each processing sub-file based on the target field of view division results. The splicing and processing module is used to control the galvanometer scanning system and motion platform to perform multi-field splicing and processing according to the processing sub-file and field of view position information, and to perform splicing seam suppression strategy during the splicing and processing process to complete the processing of large-area array structures.

[0098] This application also provides an electronic device. In some embodiments, the electronic device includes an input unit, a memory, a processor, and an output unit. The memory stores program instructions that can be executed by the processor. The processor can invoke the program instructions to execute the large-area array structure pattern segmentation and multi-view splicing processing method and / or technical solution based on the foregoing embodiments. This electronic device can be a mobile terminal device such as a mobile phone or computer.

[0099] Furthermore, embodiments of this application also provide a computer-readable storage medium for storing a computer program that performs a method for large-area array structure graphic segmentation and multi-field stitching processing. For example, computer program instructions, when executed by a computer, can invoke or provide the methods and / or technical solutions according to this application through the operation of the computer. The program instructions that invoke the methods of this application may be stored in a fixed or removable storage medium, and / or transmitted via data streams in broadcast or other signal carrying media, and / or stored in a storage medium that operates according to the program instructions.

[0100] Obviously, those skilled in the art should understand that the modules or steps of this application described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. Optionally, they can be implemented using computer-executable program code, thereby storing them in a storage device for execution by a computing device, or fabricating them separately as individual integrated circuit modules, or fabricating multiple modules or steps as a single integrated circuit module. Thus, this application is not limited to any particular combination of hardware and software.

[0101] The technical features of the above embodiments can be arbitrarily integrated. For the sake of brevity, not all possible integrations of the technical features in the above embodiments are described. However, as long as the integration of these technical features does not contradict each other, they should be considered to be within the scope of this specification.

[0102] The above embodiments merely illustrate several implementation methods of the present invention, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this invention patent should be determined by the appended claims.

Claims

1. A method for segmenting and multi-view splicing large-area array structure graphics, characterized in that, The method includes: Obtain the DXF graphic file corresponding to the array structure to be processed, parse the DXF graphic file, remove non-processable entities, perform geometric recognition on the processed entities, and extract the center coordinates and equivalent processing dimensions of the processed object; Construct a set of processing objects based on the center coordinates and equivalent processing dimensions: ； Among them, (x i ,y i Let be the center coordinates of the i-th processing object. The equivalent machining dimension of the i-th machining object; A regular grid is established based on the effective processing field of view of the galvanometer, and the initial field of view is divided for the set of processing objects to obtain multiple initial fields of view. For each processing object, establish a minimum bounding rectangle: ； Among them, O i For any object to be processed, consider it as a circular object with its center at (x). i ,y i Equivalent machining radius r i =d proc / 2,d proc The equivalent machining dimension of the object being machined; An R-tree spatial index is established based on the minimum bounding rectangle, and cross-boundary detection is performed on each initial field of view boundary to determine the cross-boundary processing object; Based on the cross-boundary detection results, perform overall translation optimization on the regular mesh: ； Among them, Ω ij For the (i, j)th grid region, Δ x and Δ y These represent the translation amounts of the grid in the x and y directions, respectively. After overall translation optimization, adaptive adjustment is performed on the local boundaries of the processing objects that still cross boundaries to obtain the target field of view division results; Based on the target field of view division results, generate multiple corresponding processing sub-files and the field of view position information corresponding to each processing sub-file; Based on the processing sub-file and field of view position information, the galvanometer scanning system and motion platform are controlled to perform multi-field stitching processing, and a stitching seam suppression strategy is implemented during the stitching process to complete the processing of large-area array structures.

2. The method for large-area array structure pattern segmentation and multi-view splicing processing as described in claim 1, characterized in that, The parsing of DXF graphics files includes: Perform coordinate unification and format standardization processing on DXF graphic files; The dimensions of the workpiece are compensated and corrected based on the preset processing compensation amount; Remove hidden layers, guide lines, annotation elements, and non-processed entities; Identify circular or block reference entities and extract the center coordinates and equivalent machining dimensions of the object to be machined.

3. The method for large-area array structure pattern segmentation and multi-view splicing processing as described in claim 2, characterized in that, The cross-boundary detection includes: Construct the boundary query region based on the field of view boundary; The R-tree spatial index is used to retrieve candidate processing objects located within the boundary query region; Perform geometric determination on candidate processing objects to identify cross-boundary processing objects that cross the field of view boundary.

4. The method for large-area array structure pattern segmentation and multi-view splicing processing as described in claim 3, characterized in that, The overall translation optimization includes: While keeping the regular mesh size constant, apply different translation amounts to the regular mesh; Count the number of cross-boundary processing objects corresponding to different translation amounts; The target translation amount that minimizes the number of cross-boundary processing objects is determined as the optimal overall translation amount.

5. The method for large-area array structure graphic segmentation and multi-view splicing processing as described in claim 4, characterized in that, The adaptive adjustment of the local boundary includes: For cross-boundary processing objects that still exist after overall translation optimization, perform dual boundary adjustments. This causes one of two adjacent fields of view to increase in size while the other field of view decreases by the same amount. Keep the total size of two adjacent fields of view unchanged, and update the corresponding field of view position information synchronously.

6. The method for large-area array structure pattern segmentation and multi-view splicing processing as described in claim 1, characterized in that, The seam suppression strategy includes at least one: Adjust the boundaries of adjacent fields of view according to the preset transition zone width; Plan the multi-field processing sequence based on the field-of-view adjacency relationship; Different processing layers are divided into layers based on different fields of view; The initial mesh position is changed between different processing cycles, causing the splicing boundary corresponding to adjacent cycles to shift.

7. The method for large-area array structure graphic segmentation and multi-view splicing processing as described in claim 6, characterized in that, The step of planning the multi-field processing sequence based on the field-of-view adjacency relationship includes: Establish field-of-view adjacency relationships based on the location information of each field of view; Divide multiple fields of view into at least two groups of non-adjacent processing fields of view; Process each group's field of view sequentially according to its group, ensuring that there is a processing time interval between adjacent fields of view.

8. A large-area array structure graphic segmentation and multi-view splicing processing device, characterized in that, The device includes: The file parsing module is used to obtain the DXF graphic file corresponding to the array structure to be processed, parse the DXF graphic file, remove non-processable entities, perform geometric recognition on the processed entities, and extract the center coordinates and equivalent processing dimensions of the processed object. The collection construction module is used to construct a collection of processing objects based on the center coordinates and equivalent processing dimensions. ； Among them, (x i ,y i Let be the center coordinates of the i-th processing object. The equivalent machining dimension of the i-th machining object; The initial division module is used to establish a regular grid based on the effective processing field of view size of the galvanometer, and to perform initial field of view division on the set of processing objects to obtain multiple initial fields of view; The rectangle creation module is used to create a minimum bounding rectangle for each processing object. ； Among them, O i For any object to be processed, consider it as a circular object with its center at (x). i ,y i Equivalent machining radius r i =d proc / 2,d proc The equivalent machining dimension of the object being machined; An R-tree spatial index is established based on the minimum bounding rectangle, and cross-boundary detection is performed on each initial field of view boundary to determine the cross-boundary processing object; The translation optimization module is used to perform overall translation optimization on the regular mesh based on the cross-boundary detection results. ； Among them, Ω ij For the (i, j)th grid region, Δ x and Δ y These represent the translation amounts of the grid in the x and y directions, respectively. The boundary adjustment module is used to perform adaptive adjustment on the local boundaries of processing objects that still cross boundaries after the overall translation optimization, so as to obtain the target field of view division result; The information generation module is used to generate multiple corresponding processing sub-files and the field of view position information of each processing sub-file based on the target field of view division results. The splicing and processing module is used to control the galvanometer scanning system and the motion platform to perform multi-field splicing and processing according to the processing sub-file and field position information, and to perform a seam suppression strategy during the splicing and processing process to complete the processing of large-area array structures.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the large-area array structure graphic segmentation and multi-field stitching processing method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the large-area array structure graphic segmentation and multi-view splicing processing method according to any one of claims 1 to 7.

Images