A splicing and drawing method for special-shaped blinds based on spatial coordinates

Abstract

The present invention discloses a method for splicing and drawing fragments of special-shaped blinds based on spatial coordinates, which relates to the field of intelligent construction. The method obtains an initial point cloud of the window surface by using a laser radar scanning device, establishes a three-dimensional point set representation in a local spatial coordinate system, and then executes a multi-level boundary point extraction process including boundary edge recognition, mutation point detection and structure segmentation to construct a candidate boundary point set Bpre; on this basis, a five-level spatial structure partitioning strategy is applied to set partition point distribution rules for different geometric characteristic areas, and finally form a window boundary point set B; the point set is highly adaptable in terms of structural continuity, spatial distribution density, sampling accuracy and boundary integrity, and significantly improves the geometric restoration degree of subsequent splicing modeling; the scheme overcomes the problem that traditional methods are difficult to deal with the local complex curvature and structural jumps of special-shaped windows, and realizes the continuous mapping of structural boundary information to splicing pattern construction.

Description

Technical Field

[0001] The present invention relates to the field of intelligent construction technology, and in particular to a spatial coordinate-based splicing and drawing method for special-shaped shutter segments. Background Art

[0002] As architectural design develops towards personalization, diversification and intelligence, special-shaped facades and irregular window structures are widely used in modern buildings, especially in high-end commercial complexes, art exhibition halls, ecological office spaces and other architectural scenes. Special-shaped shutters, as important components for building shading, aesthetic decoration and intelligent lighting adjustment, are increasingly showing significant design value and functional advantages. The shape of the shutters often needs to be spliced ​​in fragments according to the actual structure of the building facade, which puts higher requirements on the accuracy, continuity and responsiveness of the splicing pattern.

[0003] At present, in the current design practice of building curtain walls and intelligent blinds, the shading splicing patterns of irregular window surfaces often rely on manual CAD drawing or fragment arrangement based on template superposition. The design logic is mostly static rule deduction, lacks accurate modeling of the real three-dimensional shape of the window surface, and is unable to optimize shading in combination with the dynamic changes of light in spatial position. Especially in irregular, irregular or complex curvature windows, traditional methods are difficult to adapt to local morphological differences and multi-scale continuity requirements. There are often problems such as splicing patterns not matching the boundary shape, uneven shading distribution, and shading dead corners or overlapping occlusions after splicing fragments are assembled, resulting in low design efficiency and poor light control effect.

[0004] The fundamental reason for the above problems is that the traditional splicing pattern generation method fails to uniformly model three-dimensional geometric information, boundary continuity and lighting requirements. The layout of the splicing segments is based only on two-dimensional projection sketches or preset templates, and lacks the ability to perceive the complexity of the window surface structure and the dynamic response to changes in lighting conditions. Furthermore, in high-curvature turning areas or areas with drastic lighting changes, due to the lack of precise control of splicing density and segment angles, it is very easy to cause insufficient or excessive shading of local structures, thereby causing abnormal effects such as uneven indoor lighting, increased energy consumption, and decreased visual comfort during use. In severe cases, it will also cause structural failures such as accumulation of splicing segment installation errors, interference of installation structures or deformation of modules, which limits the widespread promotion of special-shaped blinds in high-end building applications and the deep integration of intelligent control. Summary of the Invention

[0005] In view of the deficiencies of the prior art, the present invention provides a method for splicing and drawing special-shaped blinds segments based on spatial coordinates, which solves the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: a method for splicing and drawing special-shaped blinds based on spatial coordinates, comprising the following steps:

[0007] S1. Perform a three-dimensional scan on the outer boundary of the special-shaped blinds to obtain the initial point cloud of the window surface. Based on the initial point cloud of the window surface, a five-level spatial structure partitioning strategy is used to optimize the point distribution and construct the window boundary point set B.

[0008] S2. Establishing a boundary curvature continuity field Ccurv based on the window boundary point set B, classifying the boundary area according to the boundary curvature continuity field Ccurv, and generating an initial splicing pattern Fbase;

[0009] S3. Calculate the splicing pattern curvature error E based on the initial splicing pattern Fbase, and set the error threshold T1 to perform preliminary comparative evaluation with the splicing pattern curvature error E;

[0010] S4. After the preliminary comparative evaluation triggers the illumination reverse mapping shading optimization, collect the standardized illumination reverse data set, calculate the comprehensive illumination action matrix Limact, solve the illumination difference mapping matrix ΔLdiff, and reversely derive the shading segment density distribution Dleaf;

[0011] S5. Map the density distribution Dleaf of the spliced ​​shading segments to a two-dimensional drawing plane through a spatial coordinate mapping function Tmap, and generate a spliced ​​drawing pattern Ffinal by fusion calculation, and output it to the execution control system to drive the splicing behavior of the special-shaped blinds segments.

[0012] Preferably, said S1 includes S11;

[0013] S11. Use a laser radar scanning device to perform non-contact three-dimensional scanning on the spatial surface of the special-shaped blinds to obtain an initial point cloud corresponding to the window surface;

[0014] The window surface initial point cloud is composed of a number of three-dimensional coordinate points (x, y, z), each of which is established in a local space coordinate system with the midpoint of the lower edge of the special-shaped shutter frame as the coordinate origin (x0, y0, z0);

[0015] Where x represents the x-axis parallel to the horizontal direction of the window, y represents the y-axis parallel to the vertical direction of the window, and z represents the z-axis perpendicular to the normal direction of the window surface.

[0016] The initial point cloud of the window surface is transmitted to the pattern construction processing unit through the wireless sensor network for subsequent boundary modeling and splicing pattern generation.

[0017] Preferably, said S1 further includes S12;

[0018] S12, performing boundary point extraction processing based on the initial point cloud of the window surface in the pattern construction processing unit to obtain a candidate boundary point set Bpre;

[0019] The boundary point extraction process includes boundary edge area recognition, mutation point recognition and structure recognition;

[0020] Based on each boundary point in the candidate boundary point set Bpre, the first-order derivative D1, second-order derivative D2 and third-order derivative D3 of all boundary points are calculated to obtain discrete derivative response values;

[0021] And according to the discrete derivative response value and the change threshold, the five-level spatial structure partitioning strategy is implemented for each boundary point in the candidate boundary point set Bpre to optimize the point layout and obtain the five-level spatial structure partition; then the corresponding partitioning point layout rules are implemented for different areas in the five-level spatial structure partition;

[0022] The five-level spatial structure partition includes an edge drastic change area, a smooth gradual change area, a seam connection area, a shading edge control area and a central uniform illumination area;

[0023] The partition point distribution rule is as follows: in the edge drastically changing area, a dense sampling point with a sampling spacing of no more than 1 cm is set by using an equal arc length subdivision algorithm; in the smooth and gradual changing area, a uniformly spaced linear interpolation sampling strategy is adopted to set a sparse sampling point column with a spacing of 5 cm; in the joint connection area, material connection nodes are calibrated based on the structure recognition results and connection closure control points are set; in the shading edge control area, spatial transition buffer points are arranged by introducing the light gradient field response value; in the central uniformly illuminated area, a triangulation grid sampling method is used for standard point distribution;

[0024] All boundary sampling points generated by the five-level spatial structure partitioning are spatially numbered, coordinates are normalized, and point columns are merged to form a window boundary point set B.

[0025] Preferably, said S2 includes S21;

[0026] S21, reading the three-dimensional coordinate information of each boundary point in the window boundary point set B, and sorting the boundary points according to the topological structure to which they belong, to form an ordered sequence of boundary points connected end to end;

[0027] A spatial local neighborhood containing multiple points before and after each boundary point in the ordered boundary point sequence is constructed as the center. The local feature vector of the geometric deformation trend of the boundary point is extracted by calculating the tangent direction change rate, normal displacement change and spacing gradient between adjacent boundary points.

[0028] The extracted local eigenvectors are traversed on the overall ordered boundary point sequence. The sliding window fitting method with smoothness constraint, multi-segment B-spline interpolation algorithm and Bezier curve construction algorithm are used to model the continuity curvature response of the boundary point set, and a boundary curvature continuity field Ccurv covering the entire boundary space is constructed.

[0029] Preferably, the S2 further includes S22;

[0030] S22. Extract the response value of each boundary point in the boundary curvature continuity field Ccurv, perform statistical normalization on all response values ​​to form a standardized curvature response data set, set a curvature continuity classification threshold interval, and set the curvature continuity classification threshold interval to include a high response threshold Thigh and a low response threshold Tlow; and execute a curvature classification rule based on the curvature continuity classification threshold interval and the response value. The specific classification content is as follows:

[0031] When the response value is greater than the high response threshold Thigh, the point is determined to be in the high curvature area;

[0032] When the response value is lower than the low response threshold Tlow, the point is determined to be in the low curvature area;

[0033] When the response value is between the low response threshold Tlow and the high response threshold Thigh, it is determined to be a transition connection area;

[0034] According to the curvature classification rule, all points in the window boundary point set B are spatially marked, and adjacent point segments with consistent markings are integrated into structural classification areas to complete the boundary classification process. The corresponding splicing segment module type is selected for each classification area, and then the initial splicing pattern Fbase is generated according to the classification area order and point layout relationship. In this case,

[0035] High curvature areas match flexible bending splicing segments;

[0036] In low-curvature areas, straight line segments are used to join segments;

[0037] The transition connection area adopts angle-adjustable structural segments.

[0038] Preferably, said S3 includes S31;

[0039] S31, extracting all the splicing fragment boundary node sequences in the initial splicing pattern Fbase to form a splicing fitting curve Cfit;

[0040] Read the theoretical boundary curvature response value corresponding to each point in the boundary curvature continuity field Ccurv as the discrete expression of the target curvature curve Cref;

[0041] The curvature deviation between the splicing fitting curve Cfit and the target curvature curve Cref at the corresponding points on the boundary curve is used as the local error indicator. The overall error calculation model is constructed by discrete interval integration to calculate the curvature error E of the output splicing pattern.

[0042] Preferably, the S3 further includes S32;

[0043] S32. Setting an error threshold T1 based on the window surface geometric complexity level and the adjustable capability of the splicing segment module. The error threshold T1 is determined by establishing an error threshold reference table including different window types, curvature complexities, and the number of boundary transitions. In the initial design phase, the corresponding target error tolerance T1 is automatically matched based on the comprehensive complexity index of the current window surface model. A preliminary comparative evaluation is performed on the splicing pattern curvature error E and the target error tolerance T1. Based on the preliminary comparative evaluation results, triggering illumination reverse mapping shading optimization is performed. The specific evaluation content is as follows;

[0044] When the splicing pattern curvature error E≤target error tolerance T1, it indicates that the current initial splicing pattern Fbase meets the design requirements in terms of structural continuity and morphological accuracy, and the illumination reverse mapping shading optimization is triggered;

[0045] When the splicing pattern curvature error E> the target error tolerance T1, it means that the current splicing pattern fails to accurately fit the target boundary curvature response and there is a risk of structural mutation and shading discontinuity. At this time, the pattern reconstruction mechanism is activated;

[0046] The startup pattern reconstruction mechanism extracts the error space distribution function to identify the error-dominant area, performs segment type replacement, connection node rearrangement and local fitting reconstruction operations, generates an updated splicing pattern Fbase', and recalculates the splicing pattern curvature error E until the tolerance threshold is met before entering the next stage of illumination reverse mapping shading optimization.

[0047] Preferably, the S4 includes S41;

[0048] S41. After preliminary comparative evaluation of the trigger illumination reverse mapping shading optimization, based on the initial splicing pattern Fbase and in combination with the spatial position relationship, start the lighting sensing device; the lighting sensing device includes a multi-angle illumination imager and a dot matrix illumination sensor array, which obtains the following multi-dimensional illumination at different time periods and azimuth angles to form a three-dimensional illumination reverse data set for the window sampling points;

[0049] The three-dimensional illumination inversion dataset includes the natural incident light intensity Iext(x, y, t) at the position (x, y) at time t, the window material reflection loss coefficient Rref at the position (x, y), and the natural light projection capability Twin of the splicing pattern at the position (x, y);

[0050] The parameters in the three-dimensional illumination inversion dataset are composed of three-dimensional coordinate points (x, y, z) in space and time dimension t as index;

[0051] The three-dimensional illumination inversion dataset is normalized by using unit dimension elimination, interval remapping and angle uniform conversion, and a unit consistency mapping function is used to normalize all sampling dimensions to the interval [0, 1] to eliminate the dimensional inconsistency interference caused by different acquisition sources and form a standardized illumination inversion dataset.

[0052] Preferably, the S4 further includes S42;

[0053] S42. Construct a comprehensive illumination intensity model for the window area based on the standardized illumination inversion dataset. The comprehensive illumination intensity model is constructed by combining and calculating the standardized illumination inversion dataset to obtain a comprehensive illumination matrix Limact(x, y, t) at the position (x, y) at time t.

[0054] Perform a one-to-one difference processing on the comprehensive illumination action matrix Limact(x, y, t) at the position (x, y) at time t and the target illumination requirement intensity Preq at the preset regional position (x, y) to obtain the illumination difference mapping matrix ΔLdiff(x, y) at the position (x, y);

[0055] The shading efficiency distribution function Beff(x, y, j) of the unit splicing fragment at different positions under the preset angle j is reversely deduced from the illumination difference mapping matrix ΔLdiff(x, y) at the position (x, y) to obtain the shading fragment density distribution Dleaf(x, y, j) at the position (x, y) under the angle j.

[0056] Preferably, the S5 includes S51;

[0057] S51. A spatial coordinate mapping function Tmap is obtained by geometrically unfolding the three-dimensional boundary structure of the irregular window surface in a local projection plane. The geometric unfolding is performed by using orthogonal projection of the principal curvature cut plane when the window surface is a regular curved surface. If the window surface has a broken line or irregular curvature change, a curvature-weighted unfolding is performed based on the boundary point set to map the three-dimensional coordinate point (x, y, z) to the two-dimensional drawing plane coordinate point (x', y').

[0058] The spatial coordinate mapping function Tmap is used to transform the density distribution Dleaf(x, y, j) of the mosaic mask fragment at the position (x, y) under the angle j defined in the three-dimensional coordinate system into the drawing coordinate system point by point, forming a two-dimensional plane density distribution function Dleaf(x', y', j). The density response information at the position (x, y) and angle j is retained as the input basis for generating the mosaic pattern.

[0059] Based on the spatial coordinate mapping function Tmap and the two-dimensional plane density distribution function Dleaf(x', y', j), a fusion calculation is performed to obtain a splicing drawing pattern Ffinal, and the splicing execution of the actual blinds segments is driven.

[0060] The present invention provides a spatial coordinate-based splicing and drawing method for special-shaped blinds. It has the following beneficial effects:

[0061] (1) This method uses a laser radar scanning device to obtain the initial point cloud of the window surface and establishes a three-dimensional point set representation in the local spatial coordinate system. It then performs a multi-level boundary point extraction process including boundary edge recognition, mutation point detection and structure segmentation to construct a candidate boundary point set Bpre. On this basis, a five-level spatial structure partitioning strategy is applied to set partition point distribution rules for different geometric characteristic areas, and finally form a window boundary point set B. This point set has high adaptability in terms of structural continuity, spatial distribution density, sampling accuracy and boundary integrity, which significantly improves the geometric restoration of subsequent splicing modeling. This scheme overcomes the problem that traditional methods are difficult to deal with the local complex curvature and structural jumps of irregular windows, realizes the continuous mapping of structural boundary information to splicing pattern construction, and ensures the accuracy of subsequent curvature modeling and the consistency of splicing adaptation.

[0062] (2) After the initial splicing pattern Fbase is established, this method quantifies the degree of fit of the splicing structure to the geometric boundary by constructing a global error calculation model between the splicing fitting curve Cfit and the target boundary curvature curve Cref. When the error meets the tolerance requirement, the illumination reverse mapping shading optimization process is further triggered to collect a multi-dimensional illumination reverse data set including the incident light intensity Iext, the material reflection loss Rref and the light transmittance Twin. After normalization, the comprehensive illumination action matrix Limact and the illumination difference mapping matrix ΔLdiff are established. Finally, by inversely combining ΔLdiff with the shading efficiency distribution function Beff at the preset angle j, the shading segment density distribution Dleaf (x, y, j) at the output position (x, y) and angle j is calculated. This process effectively realizes the adaptive design capability of reversely controlling the shading structure density based on the actual illumination demand, greatly improving the accurate response of the shading strategy to different spatial positions and lighting conditions, and solving the limitation of the existing splicing pattern that cannot dynamically respond to changes in the illumination field.

[0063] (3) This method accurately maps the shading segment density distribution function Dleaf (x, y, j) defined in the three-dimensional window surface boundary coordinate system to the two-dimensional drawing plane coordinate point (x', y'). In this mapping process, the main curvature section orthogonal projection or curvature weighted expansion strategy is adopted for different window surface types (regular or free-form surface) to ensure that the coordinate transformation retains the local shading response characteristics. In step S52, the density function after fusion calculation Tmap and Dleaf is output to achieve lossless transformation from three-dimensional structural data to two-dimensional control drawings. This output pattern not only retains the splicing angle response information, but also has the ability to directly drive the actuator to generate blinds segments, thereby realizing the integrated linkage of splicing pattern and execution control. Compared with the existing static template-based drawing method, the present invention provides an intelligent generation method with spatial perception ability, shading response regulation ability and pattern control ability, which greatly improves the automated splicing efficiency and engineering implementation ability of special-shaped blinds system. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 A schematic diagram of the steps of a spatial coordinate-based splicing drawing method for special-shaped blinds segments according to the present invention;

[0065] Figure 2 Schematic diagram of the initial point cloud of the window surface. DETAILED DESCRIPTION

[0066] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0067] Example 1: The present invention provides a method for splicing and drawing special-shaped blinds based on spatial coordinates. Figure 1 、 Figure 2 , including the following steps:

[0068] S1. Perform a three-dimensional scan on the outer boundary of the special-shaped blinds to obtain the initial point cloud of the window surface. Based on the initial point cloud of the window surface, a five-level spatial structure partitioning strategy is used to optimize the point distribution and construct the window boundary point set B.

[0069] S2. Establishing a boundary curvature continuity field Ccurv based on the window boundary point set B, classifying the boundary area according to the boundary curvature continuity field Ccurv, and generating an initial splicing pattern Fbase;

[0070] S3. Calculate the splicing pattern curvature error E based on the initial splicing pattern Fbase, and set the error threshold T1 to perform preliminary comparative evaluation with the splicing pattern curvature error E;

[0071] S4. After the preliminary comparative evaluation triggers the illumination reverse mapping shading optimization, collect the standardized illumination reverse data set, calculate the comprehensive illumination action matrix Limact, solve the illumination difference mapping matrix ΔLdiff, and reversely derive the shading segment density distribution Dleaf;

[0072] S5. Map the density distribution Dleaf of the spliced ​​shading segments to a two-dimensional drawing plane through a spatial coordinate mapping function Tmap, and generate a spliced ​​drawing pattern Ffinal by fusion calculation, and output it to the execution control system to drive the splicing behavior of the special-shaped blinds segments.

[0073] In this embodiment, this method addresses the complex boundaries of irregularly shaped windows, the multi-dimensional coupling of shading requirements, and the difficulty of automatically adapting pattern splicing. A complete technical process consisting of boundary geometry acquisition, curvature continuity modeling, structural error assessment, illumination inverse optimization, and pattern control generation is constructed. First, in S1, an initial window point cloud is acquired using a 3D scanning device such as a lidar. Based on a five-level spatial structure partitioning strategy, a differentiated optimization point distribution is performed to construct a window boundary point set B with spatial continuity and structural identification capabilities. This boundary point set is then used to establish a boundary curvature continuity field Ccurv, which models the response to the spatial geometry of the boundary. Based on this continuity field, boundary structure classification and splicing segment module matching are performed to generate an initial splicing pattern Fbase. A curvature error model is then constructed based on the relationship between the splicing fit curve and the target curvature response. The curvature error E of the splicing pattern is then calculated and preliminarily compared with a preset error threshold T1 to determine the degree of fit between the splicing pattern and the boundary fit, thereby determining whether to trigger the subsequent shading optimization phase. When the error meets the structural accuracy requirements, the illumination reverse mapping shading optimization process is triggered. A standardized illumination reverse data set, including data on natural light intensity, material reflection loss, and the light transmittance of the spliced ​​structure, is collected to construct a comprehensive illumination action matrix, Limact, and solve the illumination difference mapping matrix, ΔLdiff. Finally, the illumination difference and shading efficiency model is used to reversely generate the shading segment density distribution, Dleaf, to achieve reverse matching and precise distribution control between shading capacity and illumination requirements. Finally, in step S5, the shading segment density distribution function, Dleaf, is mapped to a two-dimensional coordinate domain by establishing a spatial coordinate mapping function, Tmap, from the three-dimensional structure to the two-dimensional drawing plane. This function is then fused with the initial splicing pattern, Fbase, to form the final splicing drawing pattern, Ffinal, which is output to the control system to drive the shutter actuator to complete the physical splicing. Through the aforementioned implementation path, the method of the present invention achieves a highly integrated process from boundary scanning, structure recognition, curvature modeling, reverse calculation of shading requirements, to pattern control output, with the following beneficial effects: First, it enables adaptive point placement and continuity control under irregular boundary conditions, improving the splicing structure's ability to fit complex window morphologies; second, it introduces a curvature error assessment and pattern reconstruction mechanism to enhance the structural accuracy and local coordination of the generated splicing pattern; third, it combines illumination reverse modeling with the calculation of the shading segment density distribution function to ensure that the generated pattern can achieve precise shading adjustment under different illumination conditions; fourth, it automatically maps the three-dimensional structure to the two-dimensional pattern through spatial coordinate transformation rules, making the splicing control more real-time, accurate, and capable of visual output. Overall, this method significantly improves the intelligence level, illumination adaptability, and engineering feasibility of the splicing drawing of irregular blinds, and has significant engineering promotion value and industrial application potential.

[0074] Example 2: Please refer to Figure 1 and Figure 2 , specifically: S1 includes S11;

[0075] S11. Use a laser radar scanning device to perform non-contact three-dimensional scanning on the spatial surface of the special-shaped blinds to obtain an initial point cloud corresponding to the window surface;

[0076] The initial point cloud of the window surface is composed of several three-dimensional coordinate points (x, y, z). Each three-dimensional coordinate point (x, y, z) is established in a local space coordinate system with the midpoint of the lower edge of the special-shaped shutter frame as the coordinate origin (x0, y0, z0);

[0077] Where x represents the x-axis parallel to the horizontal direction of the window, y represents the y-axis parallel to the vertical direction of the window, and z represents the z-axis perpendicular to the normal direction of the window surface.

[0078] The initial point cloud of the window surface is transmitted to the pattern construction processing unit through the wireless sensor network for subsequent boundary modeling and splicing pattern generation.

[0079] S1 also includes S12;

[0080] S12, performing boundary point extraction processing based on the initial point cloud of the window surface in the pattern construction processing unit to obtain a candidate boundary point set Bpre;

[0081] Boundary point extraction processing includes boundary edge area recognition, mutation point recognition and structure recognition;

[0082] Boundary edge area recognition uses a point cloud boundary recognition algorithm based on edge gradient enhancement to extract the boundary line segments of the window surface. The point cloud boundary recognition algorithm determines the boundary edge area by analyzing the point cloud density gradient and the neighborhood space change rate.

[0083] Mutation point identification uses a multi-scale curvature mutation detection method to extract mutation feature points on boundary segments. The multi-scale curvature mutation detection method identifies areas with significant curvature changes based on the local normal vector variability and gradient changes of high-order derivatives in the 3D point cloud, and extracts mutation feature points.

[0084] Structural recognition combines the layout information of the form structure components, performs logical segmentation of the boundary segments through topological analysis and structural attribute recognition, and obtains the structural recognition results;

[0085] The boundary edge area is used as the basic framework, and each line segment on the boundary segment of the window outer contour is interpolated in sections in combination with the mutation feature points identified by curvature mutation detection. Switching point marks and attribute identifiers are set within each segment according to the structure recognition results. Through point sequence reconstruction, abnormal point removal and topological connectivity verification, a candidate boundary point set Bpre with spatial continuity, structural attribute marking and boundary feature integrity is generated;

[0086] Based on each boundary point in the candidate boundary point set Bpre, the first-order derivative D1, slope change, second-order derivative D2, local curvature and third-order derivative D3, curvature change rate of all boundary points are calculated to obtain discrete derivative response values;

[0087] And according to the discrete derivative response value and the change threshold, the five-level spatial structure partitioning strategy is implemented for each boundary point in the candidate boundary point set Bpre to optimize the point layout and obtain the five-level spatial structure partition; then the corresponding partitioning point layout rules are implemented for different areas in the five-level spatial structure partition;

[0088] The five-level spatial structure divisions include the edge drastic change area, the smooth gradual change area, the joint connection area, the shading edge control area and the central uniform illumination area;

[0089] The zoning point distribution rule is to set dense sampling points with a sampling spacing of no more than 1 cm in the area with drastic edge changes using the equal arc length subdivision algorithm; in the smooth and gradual change area, a uniformly spaced linear interpolation sampling strategy is adopted to set a sparse sampling point column with a spacing of 5 cm; in the joint connection area, the material connection nodes are calibrated based on the structure recognition results and the connection closure control points are set; in the shading edge control area, the spatial transition buffer points are arranged by introducing the light gradient field response value to enhance the continuous shading ability of the shading boundary area; in the central uniform illumination area, the triangulation grid sampling method is used for standard point distribution;

[0090] All boundary sampling points generated by the five-level spatial structure partitioning are spatially numbered, coordinate-normalized, and point-series merged to form the final window boundary point set B that meets the structural adaptability and continuity control requirements. The window boundary point set B is used to subsequently establish the boundary curvature continuity field and drive the generation of the special-shaped blinds segment splicing pattern.

[0091] The window boundary point set B consists of multiple three-dimensional coordinate points, each of which meets the following conditions: the coordinates are referenced to a unified local spatial coordinate system, such as the midpoint of the lower edge of the window frame as the origin; all points are located on the structural boundary surface of the special-shaped blinds; the sampling spacing and distribution density are dynamically adjusted according to the five-level spatial structure partitioning strategy;

[0092] The specific division steps of the five-level spatial structure partition are as follows: the user sets the change threshold according to the curvature change rate, local curvature, curvature flat range and illumination gradient mutation. The change thresholds include curvature change rate threshold T1, local curvature threshold T3, curvature flat lower limit threshold T3, curvature flat upper limit threshold T4 and illumination gradient mutation threshold T5;

[0093] When the third-order derivative D3 exceeds the set mutation threshold T1, and the second-order derivative D2 is greater than the set curvature threshold T2, it is determined to be a sharp edge change area;

[0094] When the third-order derivative D3 is lower than T1 and the second-order derivative D2 is in the stable interval [T3, T4], the boundary point is determined to be a smooth and slowly changing area;

[0095] If the geometric derivatives of the boundary points change significantly, and the boundary points meet the structural connection identification conditions, they are determined to be joint connection areas.

[0096] If the illuminance gradient change rate at the boundary point exceeds the preset illuminance gradient mutation threshold T5 and the geometric curvature continuity is good, it is marked as a shading edge control area;

[0097] The remaining boundary points with relatively stable boundary responses, no structural connections or illumination changes are classified as the central uniformly illuminated area.

[0098] In this embodiment, a laser radar scanning device is used to perform non-contact three-dimensional scanning of the spatial outer contour of the special-shaped blinds, and an initial point cloud of the window surface composed of three-dimensional coordinate points (x, y, z) is collected. The point cloud is established in a unified local spatial coordinate system with the midpoint of the lower edge of the window frame as the coordinate origin (x0, y0, z0), thereby ensuring the spatial consistency and geometric comparability between different measurement points. After the initial point cloud of the window surface is transmitted to the pattern construction processing unit via the wireless sensor network, it is used in S12 to perform boundary point extraction and five-level spatial structure partitioning strategy optimization point layout operations, wherein the boundary point extraction includes obtaining the window surface contour based on the edge gradient enhancement point cloud boundary recognition algorithm, identifying the mutation feature points with significant curvature changes through the multi-scale curvature mutation detection method, and combining the window structure component layout information to perform structural recognition and logical segmentation, thereby forming a candidate boundary point set Bpre; then, a first-order derivative D is performed on each boundary point in the candidate boundary point set Bpre. 1. The discrete responses of the second-order derivative D2 and the third-order derivative D3 are calculated, and divided into five levels of spatial structure partitions, including the edge drastic change area, the smooth and gradual change area, the seam connection area, the shading edge control area, and the central uniform illumination area, based on the response value and the preset threshold. Different areas are matched with different point distribution rules. For example, the drastic change area uses equal arc length subdivision, and the central area uses triangulation distribution points. By normalizing the coordinates of the sampling points in each area, spatially numbering, and merging the point columns, a window boundary point set B with complete structure, geometric continuity, and controllable sampling density is finally formed. The implementation of this step not only achieves high-precision expression of the boundaries of complex and irregular window surfaces, but also ensures that the basic data for subsequent boundary curvature continuity modeling and pattern generation are highly expressive and adaptable, effectively improving the accuracy and adaptability of pattern structure control, reducing the risk of splicing errors caused by misjudgment of boundary curvature, and providing key technical support for the generation of splicing patterns with high fitting degree and high light control capability for irregular blinds.

[0099] Example 3: Please refer to Figure 1 , specifically: S2 includes S21;

[0100] S21, reading the three-dimensional coordinate information of each boundary point in the window boundary point set B, and sorting the boundary points according to the topological structure to which they belong, to form an ordered sequence of boundary points connected end to end;

[0101] A spatial local neighborhood containing multiple points before and after each boundary point in the ordered boundary point sequence is constructed as the center. The local feature vector of the geometric deformation trend of the boundary point is extracted by calculating the tangent direction change rate, normal displacement change and spacing gradient between adjacent boundary points.

[0102] The extracted local feature vectors are traversed on the overall ordered boundary point sequence. The sliding window fitting method with smoothness constraint, multi-segment B-spline interpolation algorithm and Bezier curve construction algorithm are used to model the continuity curvature response of the boundary point set and construct a boundary curvature continuity field Ccurv covering the entire boundary space.

[0103] The boundary curvature continuity field Ccurv is a spatial response function defined on an ordered sequence of boundary points, which can output the local geometric change rate, curvature change trend and continuity smoothness level at any boundary point.

[0104] The boundary curvature continuity field Ccurv supports local response retrieval at multiple resolutions and can serve as an important input basis for subsequent boundary region classification and splicing pattern fragment selection modules.

[0105] Finally, after the boundary curvature continuity field Ccurv is modeled, the continuity response field is structurally marked and written into the pattern data cache in spatial order for subsequent classification mapping, illumination strategy adaptive response, and pattern deformation processing modules to realize the unified mapping of boundary structure features to pattern structure control.

[0106] S2 also includes S22;

[0107] S22. Extract the response value of each boundary point in the boundary curvature continuity field Ccurv, perform statistical normalization on all response values, form a standardized curvature response data set, set a curvature continuity grading threshold interval, and set the curvature continuity grading threshold interval. The curvature continuity grading threshold interval includes a high response threshold Thigh and a low response threshold Tlow. Execute the curvature classification rule based on the curvature continuity grading threshold interval and the response value. The specific classification content is as follows:

[0108] When the response value is greater than the high response threshold Thigh, the point is determined to be in the high curvature area;

[0109] When the response value is lower than the low response threshold Tlow, the point is determined to be in the low curvature area;

[0110] When the response value is between the low response threshold Tlow and the high response threshold Thigh, it is determined to be a transition connection area;

[0111] According to the curvature classification rule, all points in the window boundary point set B are spatially marked, and adjacent point segments with consistent markings are integrated into structural classification areas to complete the boundary classification process. The corresponding splicing segment module type is selected for each classification area, and then the initial splicing pattern Fbase is generated according to the classification area order and point layout relationship. The initial splicing pattern Fbase is used to ensure that the splicing structure is consistent with the boundary curvature and supports subsequent lighting optimization design, where:

[0112] High curvature areas are matched with flexible bending splicing segments to adapt to sharply curved boundaries;

[0113] In low-curvature areas, straight segments are used to splice segments, which is suitable for straight or gently changing structures;

[0114] The transition connection area uses angle-adjustable structural segments to smoothly transition the connection relationship between different segments.

[0115] In this embodiment, the method reads all the three-dimensional coordinate point information in the window boundary point set B, and sorts them end to end according to the order of each boundary point in the spatial topological structure to construct an ordered boundary point sequence. For each point in the sequence, a spatial local neighborhood containing multiple points before and after is constructed, and local feature vectors are extracted through geometric parameters such as the tangent direction change rate, normal displacement change, and point spacing gradient, and traversal processing is performed on the entire sequence. In order to ensure the smoothness of the boundary curve and the true structural response, a sliding window fitting method with continuity constraints, multi-segment B-spline interpolation and Bezier curve construction algorithm are used to construct a boundary curvature continuity field Ccurv covering the entire boundary space. In essence, it is a spatial response function defined on the boundary point sequence. It can accurately output the local geometric change rate, curvature trend and continuous smoothness level of the point at any point, and supports multi-resolution retrieval. It is the basic basis for realizing structural intelligent adaptation and pattern module selection. Furthermore, based on the boundary curvature continuity field Ccurv, the curvature response values ​​of each boundary point are statistically normalized to form a standardized curvature response data set. A high response threshold Thigh and a low response threshold Tlow are set to form a graded threshold interval. Curvature classification rules are applied based on the relative position of the response values. All boundary points are then spatially labeled, and adjacent segments with consistent labels are integrated into structural classification regions. Based on the boundary classification, the most suitable splicing segment module type is selected for each structural region: high-curvature regions are matched with flexible bending splicing segments to accommodate sharply curved boundary morphologies; low-curvature regions are matched with straight segment splicing segments to accommodate flat or gently varying structures; and transitional regions use angle-adjustable structural segments to achieve transitional continuity between different modules. The resulting initial splicing pattern Fbase not only highly matches the actual boundary curvature continuity in terms of structural geometry, but also significantly improves the structural matching and design rationality of the splicing segments, reduces boundary abruptness issues caused by segment mismatches, and provides a core foundation for curvature fitting, lighting control, and intelligent structural splicing of special-shaped Venetian blinds.

[0116] Example 4: Please refer to Figure 1 , specifically: S3 includes S31;

[0117] S31, extracting all the splicing fragment boundary node sequences in the initial splicing pattern Fbase to form a splicing fitting curve Cfit;

[0118] Read the theoretical boundary curvature response value corresponding to each point in the boundary curvature continuity field Ccurv as the discrete expression of the target curvature curve Cref;

[0119] The curvature deviation between the splicing fitting curve Cfit and the target curvature curve Cref at the corresponding point on the boundary curve is used as the local error index. The overall error calculation model is constructed by discrete interval integration to calculate the curvature error E of the output splicing pattern.

[0120] The curvature error E of the stitching pattern is calculated and output by the following algorithm formula;

[0121] ;

[0122] Where, represents the splicing fitting curve Cfit composed of the actual splicing fragments at the i-th boundary point of the initial splicing pattern Fbase, Represents the target curvature curve Cref of the boundary curvature continuity field Ccurv at the i-th boundary point, A represents the two-dimensional window domain covered by the splicing area, the integration interval is the total splicing area, dxdy represents the integral variable on the two-dimensional window surface, and the domain integration is performed on the entire splicing area;

[0123] The overall error calculation model is a structural form with continuously differentiable parameters and can be represented by integrals. In order to evaluate the structural response difference between the initial splicing pattern Fbase and the target boundary curvature continuity field Ccurv, the mean square error index between the fitting curvature and the reference curvature is calculated on the discrete point set of the boundary to construct the splicing pattern curvature error E, so as to achieve a quantifiable evaluation of the degree of structural morphological fitting. The smaller the error value, the closer the pattern structure fits the boundary target curve and the stronger the transition continuity. 1 / A represents the normalization factor, which ensures that the error result is independent of the boundary length and is comparable. The square root operation is used to construct the RMS root mean square error, which is integrated along the entire length of the boundary to form a global continuous error surface.

[0124] S3 also includes S32;

[0125] S32, setting an error threshold T1 based on the window surface geometric complexity level and the adjustable capability of the splicing segment module. The error threshold T1 is determined by establishing an error threshold reference table (such as Table 1) that includes different window types, curvature complexities, and the number of boundary transitions. In the initial design stage, the corresponding target error tolerance T1 is automatically matched according to the comprehensive complexity index of the current window surface model.

[0126] Table 1: Error threshold reference table

[0127] Window type Curvature complexity Number of boundary turns Recommended error threshold T1 0 (rules) 0.0-0.2 0-2 0.1 0 (rules) 0.2-0.5 3-5 0.15 1 (arc) 0.3-0.6 1-4 0.2 2 (Compound) 0.5-0.8 4-7 0.3 3 (Free) 0.7-1.0 7-12 0.5

[0128] A preliminary comparative evaluation is performed on the curvature error E of the stitched pattern and the target error tolerance T1. Based on the preliminary comparative evaluation results, the triggered illumination reverse mapping shading optimization is performed. The specific evaluation contents are as follows;

[0129] When the splicing pattern curvature error E≤target error tolerance T1, it indicates that the current initial splicing pattern Fbase meets the design requirements in terms of structural continuity and morphological accuracy, and the illumination reverse mapping shading optimization is triggered;

[0130] When the splicing pattern curvature error E> the target error tolerance T1, it means that the current splicing pattern fails to accurately fit the target boundary curvature response and there is a risk of structural mutation and shading discontinuity. At this time, the pattern reconstruction mechanism is activated;

[0131] The pattern reconstruction mechanism is started to identify the error-dominant area by extracting the error space distribution function, performing fragment type replacement, connection node rearrangement and local fitting reconstruction operations to generate an updated splicing pattern Fbase' and recalculate the splicing pattern curvature error E until the tolerance threshold is met before entering the next stage of illumination reverse mapping shading optimization.

[0132] In this embodiment, the method completes the quantitative evaluation and dynamic optimization of the fitting accuracy between the initial splicing pattern Fbase and the target boundary curvature continuity field Ccurv through S31 and S32. In S31, all the splicing segment boundary nodes are first extracted from the initial splicing pattern Fbase to form a splicing fitting curve Cfit, which is then matched with the target curvature curve Cref in the boundary curvature continuity field Ccurv. By using the curvature deviation of the splicing fitting curve Cfit and the target curvature curve Cref at each point on the boundary curve as a local error indicator, a discrete interval integration method is used to construct an overall error calculation model, and the splicing pattern curvature error E is output. This curvature error E model uses a mean square error evaluation method and adds a normalization factor 1 / A to eliminate the influence of boundary size on the error result. At the same time, the RMS root mean square calculation of the square difference reflects the true continuous distribution characteristics of the error in the entire window splicing area, thereby achieving a quantitative evaluation of the degree of structural fit. The smaller the error, the more closely the splicing curve fits the target boundary and the more natural the curvature transition. S32 further establishes an error threshold reference table consisting of three elements: window type, curvature complexity, and number of boundary turning points, based on the geometric complexity level of the special-shaped window surface structure and the adjustable ability of the splicing module, through a table lookup method. The system automatically matches and sets the target error tolerance T1 during the design phase. Afterwards, the calculated splicing pattern curvature error E is preliminarily compared with the target error tolerance T1. The reconstruction mechanism automatically identifies the error-dominant area by constructing an error space distribution function, and performs operations such as fragment type replacement, node structure rearrangement, and local fitting reconstruction in the corresponding area to generate an updated splicing pattern Fbase', and then repeats the above curvature error calculation and evaluation process until the error meets the requirements. On the one hand, this implementation method ensures that the splicing pattern is highly aligned with the structural form and geometric curvature, so that the splicing boundary transition is natural and the shading effect is continuous and uniform. On the other hand, by introducing an automatic iterative reconstruction mechanism, the adaptive ability of pattern optimization is enhanced, human intervention is avoided, and the efficiency of intelligent design is improved. If the expression formula of the "stitching pattern curvature error E" is unclear, its essence is: within the stitching area A, the difference between the stitching fitting curve Cfit and the target curvature curve Cref is squared at each point and then integrated over the surface. The integral result is normalized and square rooted to obtain a scalar indicator that can reflect the accuracy of the overall stitching pattern, which is used as the core decision parameter for optimization triggering.

[0133] Example 5: Please refer to Figure 1 , specifically: S4 includes S41;

[0134] S41. After preliminary comparative evaluation and triggering of reverse mapping shading optimization, based on the initial spliced ​​pattern Fbase and in combination with spatial position relationships, activate the daylight sensing device. The daylight sensing device includes a multi-angle illumination imager and a dot matrix illumination sensor array, which acquire the following multi-dimensional illumination at different time periods and azimuth angles to form a three-dimensional illumination reverse data set for the window sampling points.

[0135] The specific acquisition method is as follows: the illumination inverse dataset is collected by a multi-dimensional illumination sensor array deployed on the outside of the window and the interior of the room. The sensor array includes multi-angle photodiode sensors, area array illumination imaging modules, or structured light scanning illumination detection units. The illumination incident simulation boundary is constructed based on the spatial pose model of the initial mosaic pattern Fbase or the updated mosaic pattern Fbase', improving the accuracy of the reverse modeling of the impact of the shading structure on the actual illumination.

[0136] The three-dimensional illumination inversion dataset includes the natural incident light intensity Iext(x, y, t) at the position (x, y) at time t, the window material reflection loss coefficient Rref at the position (x, y), and the natural light projection capability Twin of the splicing pattern at the position (x, y);

[0137] Among them, the natural incident light intensity Iext represents the natural light incident intensity, the window material reflection loss coefficient Rref represents the reflection ratio of the splicing pattern or window material to natural light, and the splicing pattern's projection ability to natural light Twin represents the light transmittance of the current splicing structure at this position, including the influence of the shielding piece spacing, material and angle;

[0138] The parameters in the three-dimensional illumination inversion dataset are composed of three-dimensional coordinate points (x, y, z) in space and time dimension t as index;

[0139] The three-dimensional illumination inversion dataset is normalized by using unit dimension elimination, interval remapping and angle uniform conversion, and a unit consistency mapping function is used to normalize all sampling dimensions to the interval [0, 1] to eliminate the dimensional inconsistency interference caused by different acquisition sources and form a standardized illumination inversion dataset.

[0140] S4 also includes S42;

[0141] S42. Construct a comprehensive illumination intensity model for the window area based on the standardized illumination inversion dataset. The comprehensive illumination intensity model obtains a comprehensive illumination matrix Limact(x, y, t) at the position (x, y) at time t by combining and calculating the standardized illumination inversion dataset.

[0142] Perform a one-to-one difference processing on the comprehensive illumination action matrix Limact(x, y, t) at the position (x, y) at time t and the target illumination requirement intensity Preq at the preset regional position (x, y) to obtain the illumination difference mapping matrix ΔLdiff(x, y) at the position (x, y) that reflects the actual illumination deviation degree. Each element of the illumination difference mapping matrix ΔLdiff is used to represent the illumination deviation degree at the position (x, y);

[0143] A positive value indicates overexposure and requires increased shading; a negative value indicates insufficient light and avoids shading.

[0144] The shading efficiency distribution function Beff(x, y, j) of the unit splicing segment at different positions under the preset angle j is reversely deduced from the illumination difference mapping matrix ΔLdiff(x, y) at the position (x, y) to obtain the density distribution Dleaf(x, y, j) of the shading segment at the position (x, y) under the angle j.

[0145] The density distribution of the light-shielded fragment Dleaf(x, y, j) at position (x, y) under angle j is calculated and output by the following algorithm formula;

[0146] ;

[0147] in, Represents the comprehensive lighting action matrix Limact(x, y, t) at the position (x, y) at time t;

[0148] Represents the lighting difference map matrix ΔLdiff(x,y) at position (x,y).

[0149] In this embodiment, when the preliminary evaluation results determine that the curvature error of the stitching pattern meets the design tolerance threshold, the system activates the spatial illumination sensing device to collect a three-dimensional illumination inversion dataset based on the initial stitching pattern Fbase or the updated stitching pattern Fbase'. The lighting sensing device includes a multi-angle illumination imager and a dot matrix illumination sensor array arranged on the inside and outside of the window surface, which can collect a three-dimensional illumination inversion dataset. The three-dimensional illumination inversion dataset has clear time t and spatial (x, y, z) indexes, and constructs the illumination simulation incident boundary by establishing a spatial pose model of the stitching pattern, so that the data collection results can more accurately reflect the actual contribution of the stitching structure to the indoor illumination. In terms of data processing, in order to avoid interference caused by differences in units and dimensions between different sampling sources, multiple normalization methods such as unit dimension elimination, interval remapping, and angle consistency conversion are adopted to form a standardized illumination inversion dataset. A comprehensive light effect intensity model is then constructed based on a standardized inverse illumination dataset. By combining and calculating the three parameters mentioned above, the comprehensive light effect matrix Limact(x, y, t) at position (x, y) at time t is obtained, accurately depicting the actual shading performance of the patchwork structure at any point. This comprehensive light effect matrix Limact(x, y, t) is then subtracted from the target plant or environment's illumination requirement Preq(x, y) at that location to form a light difference mapping matrix ΔLdiff(x, y). Positive values ​​represent overexposed areas, while negative values ​​represent underexposed areas, serving as a key indicator for shading control. Next, at each angle j, the light difference mapping matrix ΔLdiff(x, y) is reversed, combining the shading efficiency distribution function Beff(x, y, j) of the unit patch at different spatial locations. This yields the patchwork shading segment density distribution Dleaf(x, y, j), representing the density level of the patchwork segments required at a specific location and angle. This directly guides the subsequent local density construction and segment angle setting of the drawing pattern. The innovation of this implementation lies in its reverse calculation based on actual lighting conditions. This ensures that the tiled pattern not only maintains structural boundary continuity but also dynamically responds to indoor lighting requirements, resulting in a high degree of adaptability and functionality. Compared to traditional static patterns or empirical shading designs, this solution implements data-driven shading optimization, allowing the blinds to automatically generate precisely matched tiled patterns under varying lighting conditions. This significantly improves indoor light quality and energy-saving control capabilities, making it particularly suitable for shading system design in greenhouses, exhibition halls, and other spaces with high lighting requirements.

[0150] Example 6: Please refer to Figure 1 , specifically: S5 includes S51;

[0151] S51. A spatial coordinate mapping function Tmap is obtained by geometrically unfolding the three-dimensional boundary structure of the irregular window surface in a local projection plane. The geometric unfolding is performed by using orthogonal projection of the principal curvature cut plane when the window surface is a regular curved surface. If the window surface has broken lines or irregular curvature changes, a curvature-weighted unfolding is performed based on the boundary point set to map the three-dimensional coordinate point (x, y, z) to the two-dimensional drawing plane coordinate point (x', y').

[0152] The spatial coordinate mapping function Tmap is used to transform the density distribution Dleaf(x, y, j) of the mosaic mask fragment at the position (x, y) under the angle j defined in the three-dimensional coordinate system into the drawing coordinate system point by point, forming a two-dimensional plane density distribution function Dleaf(x', y', j). The density response information at the position (x, y) and angle j is retained as the input basis for generating the mosaic pattern.

[0153] Based on the fusion calculation of the spatial coordinate mapping function Tmap and the two-dimensional plane density distribution function Dleaf(x', y', j), the splicing drawing pattern Ffinal is obtained to drive the splicing execution of the actual blinds segments;

[0154] The spliced ​​drawing pattern Ffinal is calculated and output by the following algorithm formula;

[0155] ;

[0156] Where Ffinal(x', y') represents the spliced ​​drawing pattern at the coordinate point (x', y') on the two-dimensional drawing plane. It represents the function composite operation, which can be regarded as moving the value of the masked fragment density distribution Dleaf(x, y, j) at the position (x, y) under angle j to the coordinate point (x', y') on the two-dimensional drawing plane after being mapped by the spatial coordinate mapping function Tmap;

[0157] Represents the spatial coordinate mapping function Tmap from three-dimensional to two-dimensional geometric expansion transformation function;

[0158] This formula is used to project and reconstruct the shading requirement density Dleaf(x, y, j) in the three-dimensional window space onto a two-dimensional drawing plane, forming a spliced ​​drawing pattern Ffinal(x', y') that can be used for drawing printing or control signal output. Its function is to convert the spatially perceived shading control data into a two-dimensional plane pattern based on lighting differences and installation angles, which serves as the basis for subsequent pattern output and splicing control.

[0159] In this embodiment, the method maps and converts spatial shading density control data into two-dimensional pattern control information, thereby constructing a drawing pattern Ffinal based on the three-dimensional structure of the irregular blinds that can drive splicing execution. This process first geometrically unfolds the three-dimensional boundary structure of the irregular blinds within a local drawing plane to establish a spatial coordinate mapping function Tmap. The specific unfolding method is adapted based on the window geometry type: if the window surface is a regular curved surface, an orthogonal projection plane is established using the principal curvature cut planes, mapping all three-dimensional points (x, y, z) to two-dimensional drawing plane coordinates (x', y'). If the window surface has a broken line, local sudden changes, or irregular curvature, a curvature-weighted unfolding strategy is introduced, using the local curvature of each boundary point as a weighting factor. Adaptive nonlinear unfolding is performed on the three-dimensional boundary point set, ensuring that boundary continuity and density information are fully preserved on the unfolded plane. After obtaining the spatial coordinate mapping function Tmap, the previously obtained 3D shading segment density distribution Dleaf(x, y, j) at position (x, y) at angle j is mapped point by point to the 2D drawing plane, forming a 2D density distribution function Dleaf(x', y', j). This function preserves the shading density response information for each position and angle. Finally, the mapping function Tmap is fused with the 2D density function Dleaf(x', y', j) to produce the spliced ​​drawing pattern Ffinal(x', y'), which precisely projects and reconstructs the 3D shading requirements into a 2D pattern. This pattern not only contains the density information required for spatial shading control but can also be directly used for pattern output and control command generation, serving as a key control basis for subsequent printing, laser cutting, or servo drive execution. This step overcomes the technical challenge of directly applying spatial shading data to structural splicing manufacturing, completing the critical closed loop of integrated perception-computation-control pattern construction. Compared to existing manual design or two-dimensional projection approximation schemes, this method takes into account the spatial integration of multiple dependent variables, such as curvature changes, shading requirements, and installation angles. This ensures that the spliced ​​pattern meets both geometric fitting accuracy and light regulation capabilities, significantly improving the adaptive intelligent shading performance of special-shaped window components. If you have questions about the parameters: Tmap is the geometric expansion mapping function from (x, y, z) to (x', y'), which is established according to rules or weights based on the window surface structure; Dleaf(x, y, j) is the fragment density requirement function in three-dimensional space; Dleaf(x', y', j) is the result of mapping it to the drawing coordinate system, preserving the density information; and the drawing pattern Ffinal(x', y') is the final pattern output function used for splicing execution. Its core logic is a composite transformation, which transforms the density response onto the drawing plane based on the Tmap projection to form a unified structural control output.

[0160] While embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations may be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A spatial coordinate-based splicing and drawing method for special-shaped blinds, characterized by: The following steps are involved: S1. Perform a three-dimensional scan on the outer boundary of the special-shaped blinds to obtain the initial point cloud of the window surface. Based on the initial point cloud of the window surface, a five-level spatial structure partitioning strategy is used to optimize the point distribution and construct the window boundary point set B. S2. Establishing a boundary curvature continuity field Ccurv based on the window boundary point set B, classifying the boundary area according to the boundary curvature continuity field Ccurv, and generating an initial splicing pattern Fbase; S3. Calculate the splicing pattern curvature error E based on the initial splicing pattern Fbase, and set the error threshold T1 to perform preliminary comparative evaluation with the splicing pattern curvature error E; S4. After the preliminary comparative evaluation triggers the illumination reverse mapping shading optimization, collect the standardized illumination reverse data set, calculate the comprehensive illumination action matrix Limact, solve the illumination difference mapping matrix ΔLdiff, and reversely derive the shading segment density distribution Dleaf; S5. Map the density distribution Dleaf of the spliced ​​shading segments to a two-dimensional drawing plane through a spatial coordinate mapping function Tmap, and generate a spliced ​​drawing pattern Ffinal by fusion calculation, and output it to the execution control system to drive the splicing behavior of the special-shaped blinds segments.

2. The spatial coordinate-based splicing and drawing method for special-shaped Venetian blinds according to claim 1, characterized in that: Said S1 includes S11; S11. Use a laser radar scanning device to perform non-contact three-dimensional scanning on the spatial surface of the special-shaped blinds to obtain an initial point cloud corresponding to the window surface; The window surface initial point cloud is composed of a number of three-dimensional coordinate points (x, y, z), each of which is established in a local space coordinate system with the midpoint of the lower edge of the special-shaped shutter frame as the coordinate origin (x0, y0, z0); Where x represents the x-axis parallel to the horizontal direction of the window, y represents the y-axis parallel to the vertical direction of the window, and z represents the z-axis perpendicular to the normal direction of the window surface. The initial point cloud of the window surface is transmitted to the pattern construction processing unit through the wireless sensor network for subsequent boundary modeling and splicing pattern generation.

3. The spatial coordinate-based splicing and drawing method for special-shaped Venetian blinds according to claim 2, characterized in that: Said S1 also includes S12; S12, performing boundary point extraction processing based on the initial point cloud of the window surface in the pattern construction processing unit to obtain a candidate boundary point set Bpre; The boundary point extraction process includes boundary edge area recognition, mutation point recognition and structure recognition; Based on each boundary point in the candidate boundary point set Bpre, the first-order derivative D1, second-order derivative D2 and third-order derivative D3 of all boundary points are calculated to obtain discrete derivative response values; And according to the discrete derivative response value and the change threshold, the five-level spatial structure partitioning strategy is implemented for each boundary point in the candidate boundary point set Bpre to optimize the point layout and obtain the five-level spatial structure partition; then the corresponding partitioning point layout rules are implemented for different areas in the five-level spatial structure partition; The five-level spatial structure partition includes an edge drastic change area, a smooth gradual change area, a seam connection area, a shading edge control area and a central uniform illumination area; The partition point distribution rule is as follows: in the edge drastically changing area, a dense sampling point with a sampling spacing of no more than 1 cm is set by using an equal arc length subdivision algorithm; in the smooth and gradual changing area, a uniformly spaced linear interpolation sampling strategy is adopted to set a sparse sampling point column with a spacing of 5 cm; in the joint connection area, material connection nodes are calibrated based on the structure recognition results and connection closure control points are set; in the shading edge control area, spatial transition buffer points are arranged by introducing the light gradient field response value; in the central uniformly illuminated area, a triangulation grid sampling method is used for standard point distribution; All boundary sampling points generated by the five-level spatial structure partitioning are spatially numbered, coordinates are normalized, and point columns are merged to form a window boundary point set B.

4. The spatial coordinate-based splicing and drawing method for special-shaped Venetian blinds according to claim 3, characterized in that: Said S2 includes S21; S21, reading the three-dimensional coordinate information of each boundary point in the window boundary point set B, and sorting the boundary points according to the topological structure to which they belong, to form an ordered sequence of boundary points connected end to end; A spatial local neighborhood containing multiple points before and after each boundary point in the ordered boundary point sequence is constructed as the center. The local feature vector of the geometric deformation trend of the boundary point is extracted by calculating the tangent direction change rate, normal displacement change and spacing gradient between adjacent boundary points. The extracted local eigenvectors are traversed on the overall ordered boundary point sequence. The sliding window fitting method with smoothness constraint, multi-segment B-spline interpolation algorithm and Bezier curve construction algorithm are used to model the continuity curvature response of the boundary point set, and a boundary curvature continuity field Ccurv covering the entire boundary space is constructed.

5. The spatial coordinate-based splicing and drawing method for special-shaped Venetian blinds according to claim 4, characterized in that: Said S2 also includes S22; S22. Extract the response value of each boundary point in the boundary curvature continuity field Ccurv, perform statistical normalization on all response values ​​to form a standardized curvature response data set, set a curvature continuity classification threshold interval, and set the curvature continuity classification threshold interval to include a high response threshold Thigh and a low response threshold Tlow; and execute a curvature classification rule based on the curvature continuity classification threshold interval and the response value. The specific classification content is as follows: When the response value is greater than the high response threshold Thigh, the point is determined to be in the high curvature area; When the response value is lower than the low response threshold Tlow, the point is determined to be in the low curvature area; When the response value is between the low response threshold Tlow and the high response threshold Thigh, it is determined to be a transition connection area; According to the curvature classification rule, all points in the window boundary point set B are spatially marked, and adjacent point segments with consistent markings are integrated into structural classification areas to complete the boundary classification process. The corresponding splicing segment module type is selected for each classification area, and then the initial splicing pattern Fbase is generated according to the classification area order and point layout relationship. In this case, High curvature areas match flexible bending splicing segments; In low-curvature areas, straight line segments are used to join segments; The transition connection area adopts angle-adjustable structural segments.

6. The spatial coordinate-based splicing and drawing method for special-shaped Venetian blinds according to claim 4, characterized in that: Said S3 includes S31; S31, extracting all the splicing fragment boundary node sequences in the initial splicing pattern Fbase to form a splicing fitting curve Cfit; Read the theoretical boundary curvature response value corresponding to each point in the boundary curvature continuity field Ccurv as the discrete expression of the target curvature curve Cref; The curvature deviation between the splicing fitting curve Cfit and the target curvature curve Cref at the corresponding points on the boundary curve is used as the local error indicator. The overall error calculation model is constructed by discrete interval integration to calculate the curvature error E of the output splicing pattern.

7. The spatial coordinate-based splicing and drawing method for special-shaped Venetian blinds according to claim 6, characterized in that: Said S3 also includes S32; S32, setting an error threshold T1 based on the window surface geometric complexity level and the adjustable capability of the splicing segment module, wherein the error threshold T1 is obtained by establishing an error threshold reference table including different window types, curvature complexities, and the number of boundary transitions, and automatically matching the corresponding target error tolerance T1 according to the comprehensive complexity index of the current window surface model in the initial design stage; A preliminary comparative evaluation is performed on the curvature error E of the stitched pattern and the target error tolerance T1. Based on the preliminary comparative evaluation results, the triggered illumination reverse mapping shading optimization is performed. The specific evaluation contents are as follows; When the splicing pattern curvature error E≤target error tolerance T1, it indicates that the current initial splicing pattern Fbase meets the design requirements in terms of structural continuity and morphological accuracy, and the illumination reverse mapping shading optimization is triggered; When the splicing pattern curvature error E> the target error tolerance T1, it means that the current splicing pattern fails to accurately fit the target boundary curvature response, and there is a risk of structural mutation and shading discontinuity. At this time, the pattern reconstruction mechanism is activated; The startup pattern reconstruction mechanism extracts the error space distribution function to identify the error-dominant area, performs segment type replacement, connection node rearrangement and local fitting reconstruction operations, generates an updated splicing pattern Fbase', and recalculates the splicing pattern curvature error E until the tolerance threshold is met before entering the next stage of illumination reverse mapping shading optimization.

8. The spatial coordinate-based splicing and drawing method for special-shaped Venetian blinds according to claim 7, characterized in that: Said S4 includes S41; S41. After the preliminary comparative evaluation triggers the illumination reverse mapping shading optimization, based on the initial splicing pattern Fbase and in combination with the spatial position relationship, the lighting sensing device is activated; the lighting sensing device includes a multi-angle illumination imager and a dot matrix illumination sensor array, which obtains the following multi-dimensional illumination at different time periods and azimuth angles to form a three-dimensional illumination reverse data set for the window sampling points; The three-dimensional illumination inversion dataset includes the natural incident light intensity Iext(x, y, t) at the position (x, y) at time t, the window material reflection loss coefficient Rref at the position (x, y), and the natural light projection capability Twin of the splicing pattern at the position (x, y); The parameters in the three-dimensional illumination inversion dataset are composed of three-dimensional coordinate points (x, y, z) in space and time dimension t as index; The three-dimensional illumination inversion dataset is normalized by using unit dimension elimination, interval remapping and angle uniform conversion, and a unit consistency mapping function is used to normalize all sampling dimensions to the interval [0, 1] to eliminate the dimensional inconsistency interference caused by different acquisition sources and form a standardized illumination inversion dataset.

9. The spatial coordinate-based splicing and drawing method for special-shaped Venetian blinds according to claim 8, characterized in that: Said S4 also includes S42; S42. Construct a comprehensive illumination intensity model for the window area based on the standardized illumination inversion dataset. The comprehensive illumination intensity model is constructed by combining and calculating the standardized illumination inversion dataset to obtain a comprehensive illumination matrix Limact(x, y, t) at the position (x, y) at time t. Perform a one-to-one difference processing on the comprehensive illumination action matrix Limact(x, y, t) at the position (x, y) at time t and the target illumination requirement intensity Preq at the preset regional position (x, y) to obtain the illumination difference mapping matrix ΔLdiff(x, y) at the position (x, y); The shading efficiency distribution function Beff(x, y, j) of the unit splicing fragment at different positions under the preset angle j is reversely deduced from the illumination difference mapping matrix ΔLdiff(x, y) at the position (x, y) to obtain the shading fragment density distribution Dleaf(x, y, j) at the position (x, y) under the angle j.

10. The spatial coordinate-based splicing and drawing method for special-shaped Venetian blinds according to claim 9, characterized in that: Said S5 includes S51; S51. A spatial coordinate mapping function Tmap is obtained by geometrically unfolding the three-dimensional boundary structure of the irregular window surface in a local projection plane. The geometric unfolding is performed by using orthogonal projection of the principal curvature cut plane when the window surface is a regular curved surface. If the window surface has a broken line or irregular curvature change, a curvature-weighted unfolding is performed based on the boundary point set to map the three-dimensional coordinate point (x, y, z) to the two-dimensional drawing plane coordinate point (x', y'). The spatial coordinate mapping function Tmap is used to transform the density distribution Dleaf(x, y, j) of the mosaic mask fragment at the position (x, y) under the angle j defined in the three-dimensional coordinate system into the drawing coordinate system point by point, forming a two-dimensional plane density distribution function Dleaf(x', y', j). The density response information at the position (x, y) and angle j is retained as the input basis for generating the mosaic pattern. Based on the spatial coordinate mapping function Tmap and the two-dimensional plane density distribution function Dleaf(x', y', j), a fusion calculation is performed to obtain a splicing drawing pattern Ffinal, and the splicing execution of the actual blinds segments is driven.

Images