A texture fusion method and device for multi-source heterogeneous tree three-dimensional models, a terminal and a storage medium

Abstract

The application discloses a multi-source heterogeneous tree three-dimensional model texture fusion method and device, a terminal and a storage medium, and belongs to the field of three-dimensional model processing. The method comprises the following steps: acquiring a trunk part of a target tree, reconstructing a texture map of the target tree, and obtaining a trunk texture map; constructing a laser tree model of the target tree, performing texture mapping on a trunk part of the laser tree model based on the trunk texture map, and obtaining a target trunk model; acquiring a crown simulation texture map, performing texture mapping on a crown part of the laser tree model based on the crown simulation texture map, and obtaining a target crown model; acquiring a target transition area, performing boundary weighted fusion on the target transition area, and obtaining a target three-dimensional tree model. The application realizes collaborative optimization processing of the geometry and texture of the multi-source heterogeneous tree model, and can generate a tree model which is better in material performance and scene adaptability.

Description

Technical Field

[0001] This invention relates to the field of three-dimensional model processing, and in particular to a texture fusion method, apparatus, terminal and storage medium for multi-source heterogeneous tree three-dimensional models. Background Technology

[0002] Currently, in digital twin cities, environmental simulations, and film and game visualization, the core challenge lies in balancing geometric accuracy with textural realism. The two main existing technical approaches each have their limitations. Among them, the oblique photogrammetry model has the advantage of high fidelity. RGB Texture and color reproduction are good, but they are affected by viewing angle occlusion; for example, the canopy is prone to losing texture and becoming fragmented. UV Mapping leads to difficulties in cross-model reuse; while laser scanning models have millimeter-level geometric reconstruction capabilities, due to the lack of material information, they rely on physical simulation or single-view image bonding. Physical simulation has complex and distorted parameters, while single-view image bonding is inefficient and lacks realism. It can be seen that neither of the two mainstream technical approaches can independently meet the requirements for high-precision and highly compatible visualization.

[0003] It can be seen that the fragmented nature of the texture in the oblique photogrammetry model limits its application in cross-model transfer, while the laser scanning model, due to the lack of effective spectral information, often relies on artificial synthesis or single-view projection for texture generation, resulting in distortion of local appearance features.

[0004] While existing research attempts to improve consistency through multi-view texture fusion, it has failed to address the issue of cross-modal data attribute alignment, leading to texture misalignment and stretching in the geometric splicing areas of the fused model. Obvious texture seams appear at transition boundaries, disrupting the model's visual coherence.

[0005] Therefore, there is an urgent need for a texture fusion method for multi-source heterogeneous tree 3D models, which can combine the two mainstream technical approaches and solve the problem of cross-modal data attribute alignment to meet the high-fidelity visualization requirements when modeling trees.

[0006] Therefore, existing technologies still need to be improved and enhanced. Summary of the Invention

[0007] To address the aforementioned shortcomings of existing technologies, this invention provides a texture fusion method, apparatus, and terminal for multi-source heterogeneous tree 3D models, aiming to solve the problem that in existing technologies, neither of the two mainstream technical approaches can independently meet the visualization requirements of high precision and high compatibility when modeling trees in environmental simulations.

[0008] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0009] A first aspect of the present invention provides a texture fusion method for a multi-source heterogeneous tree 3D model, the method comprising:

[0010] Obtain a tilted photographic tree model of the trunk of the target tree, and reconstruct the texture map of the target tree to obtain the trunk texture map;

[0011] Construct a laser tree model of the target tree, and perform texture mapping on the trunk part of the laser tree model based on the trunk texture map to obtain the target tree trunk model;

[0012] Obtain a tree canopy simulation texture map, and perform texture mapping on the tree canopy part of the laser tree model based on the tree canopy simulation texture map to obtain the target tree canopy model;

[0013] Obtain the target transition region, which is the transition region where the trunk and crown of the laser tree model connect. Perform boundary weighted fusion on the target transition region to obtain the target three-dimensional tree model.

[0014] In one implementation, the texture map of the target tree is reconstructed, including:

[0015] Fragmented texture maps are extracted from the oblique photogrammetry tree model. These fragmented texture maps are generated by a dense matching algorithm and correspond one-to-one with the geometric model patches in the oblique photogrammetry tree model.

[0016] Obtain the three-dimensional coordinates of the geometric model facets, and based on the three-dimensional coordinates of the geometric model facets, tile and map each fragmented texture map onto a two-dimensional coordinate system to obtain the target texture map;

[0017] Based on the seamless tiling texture structure, basic texture units with periodic repeating features are extracted from the target texture map, and the basic texture units are repeatedly spliced ​​in the radial and axial directions to obtain the tree trunk texture map.

[0018] In one implementation, texture mapping is performed on the trunk portion of the laser-generated tree model based on the trunk texture map, including:

[0019] The trunk portion of the laser-generated tree model is subjected to cylindrical projection parameterization processing to extract the trunk portion of the laser-generated tree model. UV coordinate;

[0020] Based on the trunk portion of the laser-generated tree model UV The coordinates are used to render the tree trunk texture map onto the trunk portion of the laser-generated tree model.

[0021] In one implementation, obtaining the target transition region includes:

[0022] Obtain the target boundary where the target trunk model and the target crown model intersect. Extend the boundary of the target trunk model towards the crown region by the target width through a mirror symmetry operation to obtain the target transition region.

[0023] In one implementation, boundary-weighted fusion is performed on the target transition region to obtain a target 3D tree model, including:

[0024] The original texture of the target tree trunk model is mirrored and extended to the tree crown region to the target transition region to obtain the tree trunk mirror texture;

[0025] The simulated tree crown texture within the target transition area is obtained, and the tree trunk mirror texture and the simulated tree crown texture are fused by boundary weighting to obtain the target three-dimensional tree model.

[0026] In one implementation, the target 3D tree model is obtained by boundary-weighted fusion of the tree trunk mirror texture and the simulated tree crown texture, including:

[0027] Obtain each pixel in the target transition region UV Coordinate position;

[0028] Calculate the distance between each pixel and the target boundary, substitute the distance between each pixel and the target boundary into the weighted fusion formula, obtain the display color of each pixel, and obtain the target 3D tree model.

[0029] In one implementation, the weighted fusion formula is:

[0030]

[0031] in, Represents pixel coordinates. Indicates after fusion The display color of the pixels at a given location. For the tree trunk mirror texture in The display color of the pixels at a given location. For the simulated tree canopy texture in The display color of the pixels at a given location. and These represent the weight coefficients of the tree trunk mirror texture and the simulated tree crown texture, respectively. .

[0032] A second aspect of the present invention provides a texture fusion device for a multi-source heterogeneous tree 3D model, comprising:

[0033] The texture acquisition module is used to acquire the oblique photographic tree model of the trunk of the target tree, and reconstruct the texture map of the target tree to obtain the trunk texture map.

[0034] The tree trunk construction module is used to construct a laser tree model of the target tree, and to perform texture mapping on the trunk part of the laser tree model based on the tree trunk texture map to obtain the target tree trunk model;

[0035] The canopy construction module is used to obtain a canopy simulation texture map, and to perform texture mapping on the canopy part of the laser tree model based on the canopy simulation texture map to obtain the target canopy model;

[0036] The weighted fusion module is used to obtain the target transition region, which is the transition region at the connection between the trunk and crown of the laser tree model. The target transition region is subjected to boundary weighted fusion to obtain the target three-dimensional tree model.

[0037] A third aspect of the present invention provides a terminal, the terminal including a processor and a computer-readable storage medium communicatively connected to the processor, the computer-readable storage medium being adapted to store a plurality of instructions, the processor being adapted to invoke the instructions in the computer-readable storage medium to perform the steps of implementing the texture fusion method for a multi-source heterogeneous tree 3D model as described in any of the preceding claims.

[0038] In a fourth aspect, the present invention provides a computer-readable storage medium storing one or more programs that can be executed by one or more processors to implement the steps of the texture fusion method for a multi-source heterogeneous tree 3D model as described in any of the preceding claims.

[0039] Compared with existing technologies, this invention provides a texture fusion method, apparatus, and terminal for multi-source heterogeneous tree 3D models. The texture fusion method involves acquiring an oblique photogrammetric tree model of the target tree trunk, reconstructing the texture map of the target tree to obtain a trunk texture map, constructing a laser-generated tree model of the target tree, and performing texture mapping on the trunk portion of the laser-generated tree model based on the trunk texture map to obtain a target trunk model. Next, a simulated crown texture map is acquired, and texture mapping is performed on the crown portion of the laser-generated tree model based on the crown simulation texture map to obtain a target crown model. Finally, a target transition region is acquired, which is the transition area at the connection between the trunk and crown of the laser-generated tree model. Boundary-weighted fusion is performed on the target transition region to obtain the target 3D tree model. The texture fusion method for multi-source heterogeneous tree 3D models proposed in this invention solves the problem that in existing technologies, neither of the two main technical approaches for modeling trees in environmental simulations can independently meet the requirements for high-precision and high-compatibility visualization. This breakthrough breaks away from the traditional reliance on a single data source for texture mapping, providing a systematic solution for the collaborative optimization of geometry and texture in multi-source heterogeneous tree models. By deeply fusing measured textures of tilted trees with precise geometry obtained from laser-engraved trees, a tree model with superior morphology, material representation, and scene adaptability is generated. Attached Figure Description

[0040] Figure 1 A flowchart illustrating an embodiment of the texture fusion method for a multi-source heterogeneous tree 3D model provided by the present invention;

[0041] Figure 2 This is an overall technical flowchart illustrating an embodiment of the texture fusion method for a multi-source heterogeneous tree 3D model provided by the present invention.

[0042] Figure 3 A comparison of tiling textures in an embodiment of the texture fusion method for a multi-source heterogeneous tree 3D model provided by the present invention;

[0043] Figure 4 Cylindrical projection, an embodiment of the texture fusion method for multi-source heterogeneous tree 3D models provided by the present invention. UV Unfolded diagram;

[0044] Figure 5 A tree model rendering is shown as an embodiment of the texture fusion method for multi-source heterogeneous tree 3D models provided by the present invention.

[0045] Figure 6 A schematic diagram illustrating the structural principle of an embodiment of the texture fusion device for a multi-source heterogeneous tree 3D model provided by the present invention.

[0046] Figure 7A schematic diagram illustrating the principle of an embodiment of the terminal provided by the present invention. Detailed Implementation

[0047] To make the objectives, technical solutions, and effects of this invention clearer and more explicit, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0048] Those skilled in the art will understand that, unless specifically stated otherwise, the singular forms “a,” “an,” “the,” and “the” used herein may also include the plural forms. It should be further understood that the term “comprising” as used in this specification means the presence of the stated features, integers, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. It should be understood that when we say an element is “connected” or “coupled” to another element, it can be directly connected or coupled to the other element, or there may be intermediate elements. Furthermore, “connected” or “coupled” as used herein can include wireless connections or wireless coupling. The term “and / or” as used herein includes all or any units and all combinations of one or more associated listed items.

[0049] It will be understood by those skilled in the art that, unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. It should also be understood that terms such as those defined in general dictionaries should be understood to have the same meaning as in the context of the prior art, and should not be interpreted in an idealized or overly formal sense unless specifically defined as herein.

[0050] The texture fusion method for multi-source heterogeneous tree 3D models provided by this invention can be applied to terminals with computing capabilities. The terminal can execute the texture fusion method for multi-source heterogeneous tree 3D models provided by this invention to model trees in environmental simulation.

[0051] Example 1

[0052] This embodiment presents a texture fusion method for a multi-source heterogeneous 3D tree model. In this embodiment, high-fidelity consistent texture representation of the overall tree model is achieved through collaborative processing of multi-source data.

[0053] Specifically, in digital twin cities, environmental simulations, and film and game scenarios, the highly realistic visualization of trees relies heavily on the accuracy of geometric structure and material representation. Current mainstream technologies mainly rely on two types of heterogeneous data sources: oblique photogrammetry models and laser scanning models. However, both of these have significant shortcomings in the compatibility of textures and geometry.

[0054] Current mainstream technologies primarily rely on two types of heterogeneous data sources: oblique photogrammetry models and laser scanning models. Oblique photogrammetry tree models are generated through 3D reconstruction using multi-view aerial or ground imagery, and their core advantage lies in achieving high realism. RGB The model features realistic texture and color information. The laser-scanned tree model is generated through structured reconstruction based on LiDAR point cloud data, enabling high-precision capture of the tree's spatial geometry.

[0055] While oblique photogrammetry tree models can provide highly realistic texture and color information, their texture mapping has a strict one-to-one correspondence with the geometric model, requiring special processing to improve compatibility with other models. Furthermore, due to viewpoint occlusion during image acquisition, oblique models struggle to acquire complete and effective texture information in the tree canopy region. In contrast, while laser-based tree models can accurately reconstruct the geometric structure, they lack realistic texture information. Traditional methods primarily rely on physically based simulated texture mapping or single-view image-based bonding methods. The former requires extensive parameter control and yields results with significant deviations from reality, while the latter suffers from low texture realism and low processing efficiency, both failing to meet the demands of high-fidelity visualization.

[0056] In other words, the fragmented nature of the tilted model's texture limits its application in cross-model transfer, while the laser model, lacking effective spectral information, often relies on artificial synthesis or single-view projection for texture generation, leading to distortion of local appearance features. Recent research has attempted to improve consistency through multi-view texture fusion, but has failed to solve the problem of cross-modal data attribute alignment, resulting in texture misalignment and stretching in the geometric splicing area of ​​the fused model. Especially when the tree trunk model (tilted source) and the tree crown model (laser source) are spliced, due to different texture mapping mechanisms, obvious texture seams are generated at the transition boundary, disrupting the model's visual coherence.

[0057] In this embodiment, to address the aforementioned issues, taking advantage of the rich texture and color details of the tilted tree trunk model, a high-resolution seamless tiling texture map is generated using a cylindrical projection-based texture inverse mapping method. This is then combined with physically based rendering (PBR). Physically - basedRendering , PBR This technology synthesizes multiple texture maps, including parameters such as albedo, normals, and roughness, to generate high-quality tree textures suitable for 3D special effects rendering. For laser-generated tree models lacking color information, a texture coordinate system is constructed using a parametric cylindrical unfolding algorithm, adaptively mapping the tiling texture to the canopy geometry. To achieve texture continuity between heterogeneous models, a boundary-weighted fusion algorithm is introduced to eliminate seam traces, ultimately establishing a geometrically and texturally consistent 3D tree model.

[0058] Specifically, such as Figure 1 As shown, in one embodiment of the texture fusion method for a multi-source heterogeneous tree 3D model provided by the present invention, the texture fusion method for the multi-source heterogeneous tree 3D model includes the following steps:

[0059] S 100. Obtain a tilted photographic tree model of the trunk of the target tree, and reconstruct the texture map of the target tree to obtain the trunk texture map.

[0060] Reconstructing the texture map of the target tree includes:

[0061] Fragmented texture maps are extracted from the oblique photogrammetry tree model. These fragmented texture maps are generated by a dense matching algorithm and correspond one-to-one with the geometric model patches in the oblique photogrammetry tree model.

[0062] Obtain the three-dimensional coordinates of the geometric model facets, and based on the three-dimensional coordinates of the geometric model facets, tile and map each fragmented texture map onto a two-dimensional coordinate system to obtain the target texture map;

[0063] Based on a seamless tiling texture structure, basic texture units with periodic repeating features are extracted from the target texture map. The basic texture units are then repeatedly spliced ​​in the radial and axial directions to obtain the tree trunk texture map. Specifically, the tree trunk texture map is obtained by infinitely repeating the basic texture units in the radial and axial directions.

[0064] Specifically, in cutting-edge practices of digital 3D modeling and virtual reality applications, high-precision texture reconstruction of complex curved objects such as trees in natural scenes has always been a key challenge.

[0065] To achieve realistic rendering of the target tree trunk, oblique photogrammetry is typically used to acquire high-resolution multi-angle image data of the target tree. These detailed images from different perspectives form the basic material for constructing the 3D model, containing rich surface detail information. Specifically, referring to... Figure 2These high-definition photos captured from different angles are processed by a dense matching algorithm to accurately extract feature point clouds from the object's surface, thereby generating a detailed 3D tree trunk model. In this process, traditional texture mapping mechanisms directly associate color information captured from various viewpoints with corresponding geometric faces. While this ensures the visual realism of local areas, the resulting texture maps suffer from fragmentation due to over-reliance on the original geometric topology—each triangle is bound to an independent and discontinuous texture fragment. This distributed storage method not only leads to low memory usage efficiency but also limits the texture resource to specific models, preventing cross-model sharing and reuse, severely hindering workflow optimization during large-scale scene construction. To overcome this bottleneck, this embodiment innovatively proposes a cross-modal texture transfer solution based on cylindrical projection parameterization. Through mathematical transformation, the originally chaotic fragmented textures are reorganized into a spatially continuous tiling mapping pattern. This allows the processed texture maps to retain their original detailed features while being infinitely extendable and reusable like wallpaper, successfully achieving texture generalization applications between heterogeneous models. In this embodiment, fragmented texture maps generated by a dense matching algorithm are automatically extracted. These special maps are characterized by each pixel precisely corresponding to the spatial position of a specific facet of the original geometric model, forming a strict mapping relationship.

[0066] Reference Figure 3 To optimize and integrate texture resources, this embodiment employs a 3D-to-2D spatial transformation scheme. This accurately acquires the coordinate parameters of each geometric facet in 3D space, and based on this, systematically projects these scattered, fragmented textures onto a unified 2D coordinate system plane. This transformation process is akin to unfolding a 3D jigsaw puzzle into a 2D drawing, preserving the original precision while overcoming the limitations of fragmentation. After precise calculations and coordinate transformation, the texture fragments originally scattered across various surfaces are orderly reassembled into a complete target texture map.

[0067] Furthermore, this embodiment also includes identifying basic texture units with periodic repetition patterns. These meticulously designed basic pattern modules exhibit excellent tiling properties, enabling seamless, infinite extension in both the radial and axial directions. Through intelligent stitching technology, these standardized texture units are arranged and combined in an array according to preset rules, ultimately forming a reusable tree trunk texture map. This construction of a spatially continuous, tilable texture map not only perfectly solves the problem of obvious texture seams in traditional methods but also makes texture resources highly reusable, significantly improving rendering efficiency and visual consistency. The entire process represents a technological breakthrough from fragmented acquisition to systematic reconstruction, providing a reliable solution for high-precision tree modeling.

[0068] S 200. Construct a laser tree model of the target tree, and perform texture mapping on the trunk part of the laser tree model based on the trunk texture map to obtain the target tree trunk model.

[0069] Texture mapping is performed on the trunk portion of the laser-generated tree model based on the trunk texture map, including:

[0070] The trunk portion of the laser-generated tree model is subjected to cylindrical projection parameterization processing to extract the trunk portion of the laser-generated tree model. UV coordinate;

[0071] Based on the trunk portion of the laser-generated tree model UV The coordinates are used to render the tree trunk texture map onto the trunk portion of the laser-generated tree model.

[0072] Specifically, due to the unique operating characteristics of laser scanning equipment and the limitations of its underlying structured reconstruction algorithms, the generated laser tree models often only present a basic white film morphology, completely lacking the realistic textures and color variations that trees in nature should possess. This monotonous representation clearly fails to meet the demands of modern digital content creation for visual realism; therefore, an additional texture generation process must be introduced to compensate. As the core link connecting three-dimensional geometry and two-dimensional image space, the importance of texture mapping technology is self-evident—it relies on carefully designed… UV A coordinate system is used to pinpoint the precise sampling location of each vertex on the model surface within the texture image. Given that laser-reconstructed tree models commonly use cylinders as their basic geometric representation, this embodiment develops an intelligent texture unwrapping method based on cylinder projection: (Refer to...) Figure 4 By unfolding a complex three-dimensional surface into a two-dimensional shape along the generatrix direction, and combining this with parametric mapping relationships, the optimal solution is automatically calculated. UV The layout ensures that each branch receives a reasonably distributed and seamless texture coverage. This solution, specifically tailored for columnar structures, effectively solves the problem of texture loss in laser models, laying a solid foundation for subsequent realistic rendering.

[0073] Specifically, in this embodiment, a terrestrial 3D laser scanner is first used to collect comprehensive data on trees in their natural growth state. A high-speed pulsed laser beam captures massive spatial point cloud data of the tree trunk surface; these high-density data points form the digital foundation for accurately reconstructing the tree's morphology. Then, a professional texture mapping operation is performed based on a pre-prepared tree trunk texture map. Specifically, in this embodiment, the innovative technique of cylindrical projection parameterization is used to abstract the complex 3D tree trunk surface into a regular cylindrical geometry. Through mathematical transformations, each vertex is accurately extracted onto a two-dimensional plane. UVCoordinate system. This process is like tailoring a precise "geographic coordinate system" for the digitized tree trunk, ensuring that every detail in the texture map accurately corresponds to the specific location of the 3D model.

[0074] When completed UV After intelligent resolution of the coordinates, based on these mapping relationships, a meticulously designed tree trunk texture map is rendered pixel-by-pixel onto the surface of the laser-generated tree model. In other words, based on the tree trunk portion of the laser-generated tree model... UV The coordinates render the tree trunk texture map onto the trunk of the laser-generated tree model. This texture not only carries the microscopic features of real trees, such as lenticels and fissures, but also simulates the natural textural changes of bark over time through procedural generation technology. During the rendering process, advanced shader algorithms dynamically adjust light reflection parameters, ensuring that the synthesized tree trunk exhibits a texture effect completely consistent with the natural environment. Whether it's the transition between light and shadow under the dim morning light or the light and shadow layers under the setting sun, it achieves a visually indistinguishable level of realism. The final target tree trunk model retains the millimeter-level geometric precision obtained from laser scanning while possessing artistic expressiveness that surpasses photographic realism, providing a digital asset with both scientific and aesthetic value for subsequent applications in virtual scenes.

[0075] S 300. Obtain a tree canopy simulation texture map, and perform texture mapping on the tree canopy part of the laser tree model based on the tree canopy simulation texture map to obtain the target tree canopy model.

[0076] In the technical system of digital natural scene reconstruction, acquiring simulated tree canopy texture maps is the core step in achieving realistic vegetation rendering. By using procedural generation algorithms combined with leaf cluster sample images collected in the field, a texture database containing characteristics of various tree species is constructed. By adjusting leaf density distribution, color gradation patterns, and light transmission parameters, dynamic tree canopy effects that change with the seasons are simulated. When processing the canopy portion of the laser-generated tree model, a physically based lighting model is used for accurate texture mapping: the system automatically analyzes the topological structure of the canopy polygon mesh, decomposing the complex three-dimensional surface into unfoldable two-dimensional... UV In this block, a nonlinear deformation algorithm is used to ensure a natural and smooth transition of light and shadow between leaves. During the rendering process, each vertex intelligently samples texture pixels according to its spatial normal direction, so that the synthesized virtual canopy not only presents the three-dimensional layering of leaves but also maintains the transparent light effect between branches. Based on this, the simulated canopy texture map is mapped onto the canopy part of the laser tree model. The final generated target canopy model not only accurately reproduces the crown outline of a real tree but also dynamically... LODThe optimization technology enables a smooth visual transition from foreground to background, providing a highly realistic vegetation representation basis for ecological simulation in virtual environments.

[0077] S 400. Obtain the target transition region, which is the transition region at the connection between the trunk and crown of the laser tree model. Perform boundary weighted fusion on the target transition region to obtain the target three-dimensional tree model.

[0078] The acquisition of the target transition region includes:

[0079] Obtain the target boundary where the target trunk model and the target crown model intersect. Extend the boundary of the target trunk model towards the crown region by the target width through a mirror symmetry operation to obtain the target transition region.

[0080] Boundary-weighted fusion is performed on the target transition region to obtain a target 3D tree model, including:

[0081] The original texture of the target tree trunk model is mirrored and extended to the tree crown region to the target transition region to obtain the tree trunk mirror texture;

[0082] The simulated tree crown texture within the target transition area is obtained, and the tree trunk mirror texture and the simulated tree crown texture are fused by boundary weighting to obtain the target three-dimensional tree model.

[0083] The target 3D tree model is obtained by performing boundary-weighted fusion of the tree trunk mirror texture and the simulated tree crown texture, including:

[0084] Obtain each pixel in the target transition region UV Coordinate position;

[0085] Calculate the distance between each pixel and the target boundary, substitute the distance between each pixel and the target boundary into the weighted fusion formula, obtain the display color of each pixel, and obtain the target 3D tree model.

[0086] The weighted fusion formula is as follows:

[0087]

[0088] in, Represents pixel coordinates. Indicates after fusion The display color of the pixels at a given location. For the tree trunk mirror texture in The display color of the pixels at a given location. For the simulated tree canopy texture in The display color of the pixels at a given location. and These represent the weight coefficients of the tree trunk mirror texture and the simulated tree crown texture, respectively. ,and , , The boundary from the pixel to the target boundary i The distance, wherein boundary 1 is the boundary line of the tree trunk mirror texture side, and boundary 2 is the boundary line of the simulated tree crown texture side.

[0089] Specifically, in this embodiment, to achieve a highly realistic visual effect, an innovative hybrid texture mapping strategy is adopted to process the complex geometric model after fusion. This sophisticated process begins with the organic integration of multi-source data: for the trunk area, the real texture information captured by oblique photogrammetry is fully preserved. These high-precision materials originate from multi-angle image data taken on-site and are processed by a dense matching algorithm to present a delicate natural texture; while for the canopy, a mathematical transformation method based on the principle of cylindrical projection is used to unfold the texture coordinate system, and a carefully designed simulated texture map is accurately mapped onto the three-dimensional surface through an intelligent tiling algorithm. Although this unique dual-track texture scheme effectively improves the visual realism of different parts, it raises new technical challenges at the junction of the two textures—due to the difference in mapping mechanisms, obvious texture breaks often occur in the geometric splicing area, the so-called "seam effect," which seriously damages the overall coordination of the model.

[0090] To address this challenge, this embodiment employs an image boundary-based weighted fusion method to eliminate texture seams in the tree model. By assigning different weight values ​​to pixels in the texture transition area and superimposing them, pixel-level color gradation is achieved, thus resolving the issue of noticeable texture breaks often appearing in geometric splicing areas. Specifically, in this embodiment, the target boundary is located—the geometric contour line at the intersection of the target trunk model and the crown model is automatically detected by an algorithm. This dynamically generated digital boundary line, i.e., the target boundary, becomes the benchmark for subsequent operations. When performing mirror-symmetric expansion based on the target boundary, the influence range of the trunk texture is extended towards the crown direction with the target width, thereby creating a special transition zone with bidirectional compatibility, resulting in the target transition region.

[0091] Specifically, constructing realistic tree models requires overcoming the limitations of traditional single-texture mapping. To achieve a natural connection between the trunk and crown in the laser-generated tree model, this embodiment innovatively employs boundary-weighted fusion technology to process key transition areas. When performing texture extension operations, the original trunk texture is not simply copied linearly, but rather extended to the crown area after mirror transformation to form a mirrored texture copy. This symmetrical processing method not only preserves the realistic texture characteristics of the trunk bark but also provides a structured data foundation for subsequent hybrid calculations. Simultaneously, the system synchronously extracts the original crown texture information within the transition area; these pixel data carrying details of branches and leaves will participate in the fusion calculation together with the mirrored texture.

[0092] Specifically, the core weighted fusion algorithm is based on precise... UV Within this coordinate system, each pixel in the transition zone is assigned a unique two-dimensional coordinate location. By calculating the Euclidean distance from each pixel to the target boundary, the system constructs a gradient weight field based on spatial location. Under the influence of the fusion formula, pixels near the trunk primarily exhibit the color characteristics of a mirrored texture, while the portion closer to the canopy gradually enhances the intensity of the simulated canopy texture. This nonlinear blending strategy achieves a smooth transition between the two heterogeneous textures at the pixel level, effectively eliminating the abrupt seams produced by traditional stitching methods.

[0093] Reference Figure 5 The resulting 3D tree model exhibits a seamless visual effect: from the thick trunk to the dense branches, the texture changes follow the laws of natural growth, preserving the geometric accuracy obtained from laser scanning while incorporating procedurally generated ecological details. This technological breakthrough not only solves the problem of multi-source data fusion but also provides a reusable standardized solution for vegetation rendering in virtual scenes, enabling each tree in the digital forest to present a unique sense of realism and vitality.

[0094] Specifically, within the target transition region, two sets of texture information from different sources coexist and interact. In this embodiment, through a strategy of dynamically allocating pixel weight coefficients, the computer automatically adjusts the contribution ratio of the two textures based on their distance from the baseline, prioritizing the preservation of the original texture of the target tree trunk model near the baseline. The realistic texture gradually increases towards the outer edges, creating a more pronounced simulated tree canopy texture. The performance intensity is enhanced. This distance-field-based progressive blending algorithm achieves pixel-level color interpolation calculation, resulting in a smooth and continuous gradient effect for hue changes in transition regions, effectively eliminating harsh visual breaks.

[0095] Specifically, within the target transition area, there are two texture mapping systems: one is a mirror texture extending from the tilted tree model, namely the tree trunk mirror texture. Secondly, through UV Expand the tiled texture of the index, which is the simulated tree canopy texture. Therefore, the model has texture overlap regions at the transition points. Within these overlap regions, let... and These are the tree trunk mirror texture and the simulated tree crown texture, respectively. This indicates the merged, spliced ​​texture. This represents the coordinates of pixels within the overlapping region. Therefore, at this point, the coordinates... Texture at the location :

[0096] ;

[0097] in, The coordinates of the tree trunk boundary are... This is the target transition region. The tree trunk mirror texture is formed within the target transition region. With the simulated tree canopy texture The double coverage. Specifically, for the geometric seam area, the weighted fusion formula is:

[0098]

[0099] in, I This indicates the merged, spliced ​​texture. Represents pixel coordinates. Indicates after fusion The display color of the pixels at a given location. For the tree trunk mirror texture in The display color of the pixels at a given location. For the simulated tree canopy texture in The display color of the pixels at a given location. and These represent the weight coefficients of the tree trunk mirror texture and the simulated tree crown texture, respectively. ,and , , The boundary from the pixel to the target boundary i The distance, wherein boundary 1 is the boundary line of the tree trunk mirror texture side, and boundary 2 is the boundary line of the simulated tree crown texture side.

[0100] As can be seen, the entire fusion process resembles digital alchemy, transforming two heterogeneous textures into a seamless, organic whole through mathematical formulas. The color value of each pixel is the result of multiple factors: the color information of the original texture, the weighting coefficients determined by its spatial location, and the reflectivity under ambient lighting conditions all participate in the calculation. This physically based rendering method ensures that the visual effect of the transition area remains highly consistent regardless of the viewing angle. The final 3D tree model not only accurately replicates the growth form of natural trees in structure but also achieves a remarkably realistic artistic effect in terms of material representation, providing a groundbreaking visual solution for ecological simulation in virtual scenes.

[0101] The key aspect of this embodiment is addressing the texture consistency mapping problem of multi-source heterogeneous tree models. A cross-modal texture fusion method based on cylindrical projection is proposed, achieving seamless texture integration between the tilted tree model and the laser-generated tree model through the following core technologies: First, for texture generation of the tilted tree trunk model, a cylindrical surface is designed... UV The process involves a three-tiered workflow: coordinate unrolling, texture inverse reconstruction, and visual perception optimization. Curvature adaptive parameterization maps the 3D cylindrical geometry to a 2D model. UV A coordinate system is used to generate high-fidelity tiled texture maps containing microscopic details such as bark cracks. Secondly, to address the texture loss problem in laser-generated tree canopy models, a parameterized approach based on cylindrical geometric features is proposed. UV Unfolding method. The surface of the laser-guided tree mesh model is converted into two dimensions through multi-segment cylindrical fitting. UV A coordinate system is used to accurately index the tiling texture of tilted trees. The texture fusion method for multi-source heterogeneous tree 3D models described in this embodiment effectively solves the problem of difficult texture mapping for complex canopy structures, ensuring complete texture coverage in the canopy region. Finally, in the geometric splicing area, a distance-weighted gradient fusion algorithm is introduced. By using texture mirroring cloning technology to extend the tilted tree texture to the seam area, and utilizing a smooth weight allocation strategy, a pixel-level seamless transition between the tilted tree texture and the laser-etched tree texture is achieved.

[0102] In summary, this embodiment provides a texture fusion method for multi-source heterogeneous tree 3D models. It involves acquiring an oblique photogrammetric tree model of the target tree trunk, reconstructing the texture map of the target tree to obtain a trunk texture map, constructing a laser-generated tree model of the target tree, and performing texture mapping on the trunk portion of the laser-generated tree model based on the trunk texture map to obtain the target trunk model. Next, it acquires a simulated crown texture map and performs texture mapping on the crown portion of the laser-generated tree model based on the crown simulation texture map to obtain the target crown model. Finally, it acquires a target transition region, which is the transition area at the connection between the trunk and crown of the laser-generated tree model, and performs boundary-weighted fusion on the target transition region to obtain the target 3D tree model. The texture fusion method for multi-source heterogeneous tree 3D models proposed in this embodiment solves the problem that in existing technologies, neither of the two main technical approaches for modeling trees in environmental simulations can independently meet the visualization requirements of high precision and high compatibility. This breakthrough breaks away from the traditional reliance on a single data source for texture mapping, providing a systematic solution for the collaborative optimization of geometry and texture in multi-source heterogeneous tree models. Through deep fusion of measured textures of tilted trees and precise geometry of laser-generated trees, the generated tree model achieves a high level in terms of morphology, material representation, and scene adaptability.

[0103] It should be understood that although the steps in the flowcharts shown in the accompanying drawings are displayed sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowchart may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least a portion of the sub-steps or stages of other steps.

[0104] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided by this invention can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM). ROM Programmable ROM ( PROM ), electrically programmable ROM ( EPROM Electrically erasable programmable ROM ( EEPROM ) or flash memory. Volatile memory may include random access memory (RAM) RAM Alternatively, an external cache memory. This is for illustrative purposes only and not as a limitation. RAM It can be obtained in various forms, such as static RAM ( SRAM ),dynamic RAM ( DRAM ),synchronous DRAM ( SDRAM ), double data rate SDRAM ( DDR SDRAM ), Enhanced SDRAM ( ESDRAM ), Synchronization Link ( Synchlink ), DRAM ( SLDRAM ), memory bus ( Rambus )direct RAM ( RDRAM ), Direct Memory Bus Dynamics RAM ( DRDRAM ), and memory bus dynamics RAM ( RDRAM )wait.

[0105] Example 2

[0106] Based on the above embodiments, the present invention also provides a texture fusion device for a multi-source heterogeneous tree 3D model, such as... Figure 6 As shown, the texture fusion device for the multi-source heterogeneous tree 3D model includes:

[0107] The texture acquisition module is used to acquire the oblique photographic tree model of the trunk of the target tree, reconstruct the texture map of the target tree to obtain the trunk texture map, as described in Embodiment 1.

[0108] The tree trunk construction module is used to construct a laser tree model of the target tree. Based on the tree trunk texture map, the trunk part of the laser tree model is texture mapped to obtain the target tree trunk model, as specifically described in Embodiment 1.

[0109] The canopy construction module is used to obtain a canopy simulation texture map, and to perform texture mapping on the canopy part of the laser tree model based on the canopy simulation texture map to obtain the target canopy model, as specifically described in Embodiment 1;

[0110] The weighted fusion module is used to obtain the target transition region, which is the transition region at the connection between the trunk and crown of the laser tree model. The target transition region is subjected to boundary weighted fusion to obtain the target three-dimensional tree model, as described in Embodiment 1.

[0111] Example 3

[0112] Based on the above embodiments, the present invention also provides a terminal, such as... Figure 7 As shown, the terminal includes a processor 10 and a memory 20. Figure 7 Only some of the terminal components are shown; however, it should be understood that it is not required to implement all of the components shown, and more or fewer components may be implemented instead.

[0113] In some embodiments, the memory 20 may be an internal storage unit of the terminal, such as the terminal's hard drive or memory. In other embodiments, the memory 20 may also be an external storage device of the terminal, such as a plug-in hard drive or smart memory card equipped on the terminal. SmartMediaCard , SMC ), Secure Digital ( SecureDigital , SD ) card, flash memory card ( FlashCard Furthermore, the memory 20 may include both internal storage units and external storage devices of the terminal. The memory 20 is used to store application software and various types of data installed on the terminal. The memory 20 can also be used to temporarily store data that has been output or will be output.

[0114] In one embodiment, the memory 20 stores a texture fusion program 30 for a multi-source heterogeneous tree 3D model. The texture fusion program 30 for the multi-source heterogeneous tree 3D model can be executed by the processor 10 to realize the texture fusion method for the multi-source heterogeneous tree 3D model in this application.

[0115] In some embodiments, the processor 10 may be a central processing unit (CPU). Central Processing Unit , CPU (a microprocessor or other chip) is used to run program code stored in the memory 20 or process data, such as executing the texture fusion method of the multi-source heterogeneous tree 3D model.

[0116] In one embodiment, when the processor 10 executes the texture fusion program 30 of the multi-source heterogeneous tree 3D model in the memory 20, the following steps are performed:

[0117] Obtain a tilted photographic tree model of the trunk of the target tree, and reconstruct the texture map of the target tree to obtain the trunk texture map;

[0118] Construct a laser tree model of the target tree, and perform texture mapping on the trunk part of the laser tree model based on the trunk texture map to obtain the target tree trunk model;

[0119] Obtain a tree canopy simulation texture map, and perform texture mapping on the tree canopy part of the laser tree model based on the tree canopy simulation texture map to obtain the target tree canopy model;

[0120] Obtain the target transition region, which is the transition region where the trunk and crown of the laser tree model connect. Perform boundary weighted fusion on the target transition region to obtain the target three-dimensional tree model.

[0121] In one implementation, the texture map of the target tree is reconstructed, including:

[0122] Fragmented texture maps are extracted from the oblique photogrammetry tree model. These fragmented texture maps are generated by a dense matching algorithm and correspond one-to-one with the geometric model patches in the oblique photogrammetry tree model.

[0123] Obtain the three-dimensional coordinates of the geometric model facets, and based on the three-dimensional coordinates of the geometric model facets, tile and map each fragmented texture map onto a two-dimensional coordinate system to obtain the target texture map;

[0124] Based on the seamless tiling texture structure, basic texture units with periodic repeating features are extracted from the target texture map, and the basic texture units are repeatedly spliced ​​in the radial and axial directions to obtain the tree trunk texture map.

[0125] In one implementation, texture mapping is performed on the trunk portion of the laser-generated tree model based on the trunk texture map, including:

[0126] The trunk portion of the laser-generated tree model is subjected to cylindrical projection parameterization processing to extract the trunk portion of the laser-generated tree model. UV coordinate;

[0127] Based on the trunk portion of the laser-generated tree model UV The coordinates are used to render the tree trunk texture map onto the trunk portion of the laser-generated tree model.

[0128] In one implementation, obtaining the target transition region includes:

[0129] Obtain the target boundary where the target trunk model and the target crown model intersect. Extend the boundary of the target trunk model towards the crown region by the target width through a mirror symmetry operation to obtain the target transition region.

[0130] In one implementation, boundary-weighted fusion is performed on the target transition region to obtain a target 3D tree model, including:

[0131] The original texture of the target tree trunk model is mirrored and extended to the tree crown region to the target transition region to obtain the tree trunk mirror texture;

[0132] The simulated tree crown texture within the target transition area is obtained, and the tree trunk mirror texture and the simulated tree crown texture are fused by boundary weighting to obtain the target three-dimensional tree model.

[0133] In one implementation, the target 3D tree model is obtained by boundary-weighted fusion of the tree trunk mirror texture and the simulated tree crown texture, including:

[0134] Obtain each pixel in the target transition region UV Coordinate position;

[0135] Calculate the distance between each pixel and the target boundary, substitute the distance between each pixel and the target boundary into the weighted fusion formula, obtain the display color of each pixel, and obtain the target 3D tree model.

[0136] In one implementation, the weighted fusion formula is:

[0137]

[0138] in, Represents pixel coordinates. Indicates after fusion The display color of the pixels at a given location. For the tree trunk mirror texture in The display color of the pixels at a given location. For the simulated tree canopy texture in The display color of the pixels at a given location. and These represent the weight coefficients of the tree trunk mirror texture and the simulated tree crown texture, respectively. .

[0139] Example 4

[0140] The present invention also provides a computer-readable storage medium storing one or more programs that can be executed by one or more processors to implement the steps of the texture fusion method for a multi-source heterogeneous tree 3D model as described above.

[0141] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A texture fusion method for a multi-source heterogeneous tree 3D model, characterized in that, The texture fusion method for the multi-source heterogeneous tree 3D model includes: Obtain a tilted photographic tree model of the trunk of the target tree, and reconstruct the texture map of the target tree to obtain the trunk texture map; Construct a laser tree model of the target tree, and perform texture mapping on the trunk part of the laser tree model based on the trunk texture map to obtain the target tree trunk model; Obtain a tree canopy simulation texture map, and perform texture mapping on the tree canopy part of the laser tree model based on the tree canopy simulation texture map to obtain the target tree canopy model; Obtain the target transition region, which is the transition region where the trunk and crown of the laser tree model connect. Perform boundary weighted fusion on the target transition region to obtain the target three-dimensional tree model. Texture mapping is performed on the trunk portion of the laser-generated tree model based on the trunk texture map, including: The trunk portion of the laser-generated tree model is subjected to cylindrical projection parameterization processing to extract the trunk portion of the laser-generated tree model. UV coordinate; Based on the trunk portion of the laser-generated tree model UV The coordinates are used to render the tree trunk texture map onto the trunk portion of the laser-generated tree model; The acquisition of the target transition region includes: Obtain the target boundary where the target trunk model and the target crown model intersect. Extend the boundary of the target trunk model towards the crown region by the target width through a mirror symmetry operation to obtain the target transition region.

2. The texture fusion method for multi-source heterogeneous tree 3D models according to claim 1, characterized in that, Reconstructing the texture map of the target tree includes: Fragmented texture maps are extracted from the oblique photogrammetry tree model. These fragmented texture maps are generated by a dense matching algorithm and correspond one-to-one with the geometric model patches in the oblique photogrammetry tree model. Obtain the three-dimensional coordinates of the geometric model facets, and based on the three-dimensional coordinates of the geometric model facets, tile and map each fragmented texture map onto a two-dimensional coordinate system to obtain the target texture map; Based on the seamless tiling texture structure, basic texture units with periodic repeating features are extracted from the target texture map, and the basic texture units are repeatedly spliced ​​in the radial and axial directions to obtain the tree trunk texture map.

3. The texture fusion method for multi-source heterogeneous tree 3D models according to claim 1, characterized in that, Boundary-weighted fusion is performed on the target transition region to obtain a target 3D tree model, including: The original texture of the target tree trunk model is mirrored and extended to the tree crown region to the target transition region to obtain the tree trunk mirror texture; The simulated tree crown texture within the target transition area is obtained, and the tree trunk mirror texture and the simulated tree crown texture are fused by boundary weighting to obtain the target three-dimensional tree model.

4. The texture fusion method for a multi-source heterogeneous tree 3D model according to claim 3, characterized in that, The target 3D tree model is obtained by performing boundary-weighted fusion of the tree trunk mirror texture and the simulated tree crown texture, including: Obtain each pixel in the target transition region. UV Coordinate position; Calculate the distance between each pixel and the target boundary, substitute the distance between each pixel and the target boundary into the weighted fusion formula, obtain the display color of each pixel, and obtain the target 3D tree model.

5. The texture fusion method for a multi-source heterogeneous tree 3D model according to claim 4, characterized in that, The weighted fusion formula is as follows: in, Represents pixel coordinates. Indicates after fusion The display color of the pixels at a given location. For the tree trunk mirror texture in The display color of the pixels at a given location. For the simulated tree canopy texture in The display color of the pixels at a given location. and These represent the weight coefficients of the tree trunk mirror texture and the simulated tree crown texture, respectively. .

6. A texture fusion device for a multi-source heterogeneous tree 3D model, characterized in that, include: The texture acquisition module is used to acquire the oblique photographic tree model of the trunk of the target tree, and reconstruct the texture map of the target tree to obtain the trunk texture map. The tree trunk construction module is used to construct a laser tree model of the target tree, and to perform texture mapping on the trunk part of the laser tree model based on the tree trunk texture map to obtain the target tree trunk model; The canopy construction module is used to obtain a canopy simulation texture map, and to perform texture mapping on the canopy part of the laser tree model based on the canopy simulation texture map to obtain the target canopy model; The weighted fusion module is used to obtain the target transition region, which is the transition region at the connection between the trunk and crown of the laser tree model. The target transition region is subjected to boundary weighted fusion to obtain the target three-dimensional tree model. Texture mapping is performed on the trunk portion of the laser-generated tree model based on the trunk texture map, including: The trunk portion of the laser-generated tree model is subjected to cylindrical projection parameterization processing to extract the trunk portion of the laser-generated tree model. UV coordinate; Based on the trunk portion of the laser-generated tree model UV The coordinates are used to render the tree trunk texture map onto the trunk portion of the laser-generated tree model; The acquisition of the target transition region includes: Obtain the target boundary where the target trunk model and the target crown model intersect. Extend the boundary of the target trunk model towards the crown region by the target width through a mirror symmetry operation to obtain the target transition region.

7. A terminal, characterized in that, The terminal includes: a processor and a computer-readable storage medium communicatively connected to the processor, the computer-readable storage medium being adapted to store a plurality of instructions, and the processor being adapted to invoke the instructions in the computer-readable storage medium to execute the steps of implementing the texture fusion method for the multi-source heterogeneous tree 3D model according to any one of claims 1-5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores one or more programs, which can be executed by one or more processors to implement the steps of the texture fusion method for a multi-source heterogeneous tree 3D model as described in any one of claims 1-5.

Images