Texture generation method and device of medical image model, imaging equipment and medium

Abstract

The invention relates to the technical field of image processing, and discloses a texture generation method and device of a medical image model, imaging equipment and a medium. The method comprises the following steps: firstly, carrying out model construction on a target object to obtain an initial three-dimensional grid model and a mapping three-dimensional grid model containing color texture information; then, based on a plurality of vertexes of the initial three-dimensional grid model, mapping the initial three-dimensional grid model to a two-dimensional plane to obtain an initial map corresponding to the initial three-dimensional grid model; then, based on the three-dimensional map grid model, color assignment is carried out on the initial map, and a color map is generated; and finally, applying a color map to the initial three-dimensional grid model, and obtaining a target three-dimensional grid model containing color texture information, which is similar to an actual result, has a high sense of reality and accurately reflects texture features of tissues, so that the visual effect of the three-dimensional grid model is improved. More accurate and more visual reference is provided for clinical diagnosis and operation planning, and the diagnosis and treatment effects can be improved.

Description

Technical Field

[0001] The present invention relates to the field of image processing technology, and in particular to a texture generation method, device, imaging equipment and medium for a medical image model. Background Art

[0002] Three-dimensional reconstruction technology is widely used in medical diagnosis, but the resulting three-dimensional mesh model usually appears grayish white, lacks realism and details, and is not conducive to observing the characteristics of different tissues and lesions in the three-dimensional mesh model.

[0003] In the related art, colors are manually or automatically assigned to three-dimensional mesh models, and colors are assigned to different tissues in the reconstructed model by adjusting the material. However, the manual or automatic color assignment method is based on subjective judgment, and the same color is assigned to the same organ tissue, which cannot reflect the true texture details of the tissue. Therefore, it is necessary to propose a texture generation method for medical image models. Summary of the invention

[0004] The embodiments of this specification are intended to solve at least one of the technical problems in the related art to a certain extent. To this end, the embodiments of this specification propose a texture generation method, device, imaging device and medium for a medical image model.

[0005] The present specification provides a method for generating texture of a medical image model, the method comprising:

[0006] Acquire an initial three-dimensional mesh model and a mapped three-dimensional mesh model containing color and texture information; wherein the initial three-dimensional mesh model and the mapped three-dimensional mesh model are both obtained by model construction for the target object;

[0007] Based on multiple vertices of the initial three-dimensional mesh model, mapping the initial three-dimensional mesh model to a two-dimensional plane to obtain an initial map corresponding to the initial three-dimensional mesh model;

[0008] Based on the texture three-dimensional grid model, assigning colors to the initial texture to generate a color texture;

[0009] The color map is applied to the initial three-dimensional mesh model to obtain a target three-dimensional mesh model containing color texture information.

[0010] In one embodiment, the mapping three-dimensional mesh model is determined by:

[0011] Rendering a sequence of medical images including the target object to generate a rendering model including color texture information;

[0012] Acquire images corresponding to the rendering model from different angles to form a rendering image sequence;

[0013] Mesh reconstruction is performed based on the rendered image sequence to generate the mapped three-dimensional mesh model.

[0014] In one embodiment, the mesh reconstruction based on the rendered image sequence to generate the textured three-dimensional mesh model includes:

[0015] Performing motion analysis and conversion based on the rendered image sequence to obtain a depth image sequence corresponding to the rendered image sequence;

[0016] Dense reconstruction is performed based on the depth image sequence to generate the mapped three-dimensional mesh model.

[0017] In one embodiment, the step of assigning colors to the initial texture based on the texture three-dimensional grid model to generate a color texture includes:

[0018] Using the initial three-dimensional mesh model as a reference, aligning the mapped three-dimensional mesh model to obtain a registered three-dimensional mesh model;

[0019] The registered three-dimensional mesh model is used to assign colors to the initial map to generate the color map.

[0020] In one embodiment, the step of assigning colors to the initial map using the registered three-dimensional mesh model to generate the color map includes:

[0021] On the registered three-dimensional mesh model, determining registration vertices corresponding to vertices of the initial three-dimensional mesh model;

[0022] Using the registration vertices to assign colors to corresponding vertices of the initial three-dimensional mesh model;

[0023] Based on the mapping relationship between the initial three-dimensional mesh model and the initial map, the initial map is assigned colors using the vertices of the initial three-dimensional mesh model after color assignment to generate the color map.

[0024] In one embodiment, the method further comprises:

[0025] Generating a normal vector texture map corresponding to the initial three-dimensional mesh model;

[0026] The normal vector texture map is applied to the target three-dimensional mesh model so that the target three-dimensional mesh model has concave-convex texture information.

[0027] In one embodiment, generating a normal vector texture map corresponding to the initial three-dimensional mesh model includes:

[0028] Converting the color map to a grayscale map;

[0029] The normal vector texture map is generated based on the gradient information of the grayscale map.

[0030] In one embodiment, the initial three-dimensional mesh model is determined by performing surface rendering modeling on a medical image sequence including the target object to generate an initial three-dimensional mesh model.

[0031] In one embodiment, mapping the initial three-dimensional mesh model to a two-dimensional plane based on multiple vertices of the initial three-dimensional mesh model to obtain an initial map corresponding to the initial three-dimensional mesh model includes:

[0032] Based on a plurality of vertices of the initial three-dimensional mesh model, the surface of the three-dimensional model is divided into a plurality of segments, and each segment is mapped onto a two-dimensional plane to obtain the initial map.

[0033] The present specification provides a texture generation device for a medical image model, the device comprising:

[0034] A mesh model acquisition module, used to acquire an initial three-dimensional mesh model and a textured three-dimensional mesh model containing color and texture information; wherein the initial three-dimensional mesh model and the textured three-dimensional mesh model are both obtained by model construction for a target object;

[0035] An initial map generation module, used for mapping the initial three-dimensional mesh model to a two-dimensional plane based on a plurality of vertices of the initial three-dimensional mesh model to obtain an initial map corresponding to the initial three-dimensional mesh model;

[0036] A color map generation module, used to assign colors to the initial map based on the map three-dimensional grid model to generate a color map;

[0037] The target mesh model generation module is used to apply the color map to the initial three-dimensional mesh model to obtain a target three-dimensional mesh model containing color texture information.

[0038] An embodiment of the present specification provides a medical imaging device, which includes: a memory, and one or more processors communicatively connected to the memory; the memory stores instructions executable by the one or more processors, and the instructions are executed by the one or more processors to enable the one or more processors to implement the steps of the method described in any of the above embodiments.

[0039] The embodiments of this specification provide a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of the method described in any of the above embodiments are implemented.

[0040] An embodiment of the present specification provides a computer program product, wherein the computer program product includes instructions, and when the instructions are executed by a processor of a computer device, the computer device is enabled to perform the steps of the method described in any one of the above embodiments.

[0041] In the above-mentioned implementation mode of the specification, a model is first constructed for the target object to obtain an initial three-dimensional mesh model and a mapping three-dimensional mesh model containing color texture information. Then, based on multiple vertices of the initial three-dimensional mesh model, the initial three-dimensional mesh model is mapped to a two-dimensional plane to obtain an initial map corresponding to the initial three-dimensional mesh model. Then, based on the mapping three-dimensional mesh model, the initial map is assigned a color value to generate a color map. Finally, the color map is applied to the initial three-dimensional mesh model to obtain a target three-dimensional mesh model containing color texture information.

[0042] The above process completes the reconstruction, mapping and expansion, color map generation and other tasks of the three-dimensional mesh model through automated processing, thereby improving processing efficiency and reducing the need for manual intervention. By applying the color map to the initial three-dimensional mesh model, the target three-dimensional mesh model containing color texture information is similar to the actual result, has a high sense of reality, and accurately reflects the texture characteristics of the tissue, thereby improving the visual effect and expressiveness of the three-dimensional mesh model, providing a more accurate and intuitive reference for clinical diagnosis and surgical planning, and helping to improve the diagnosis and treatment effects. Compared with the method of manually drawing texture maps, this embodiment does not require the participation of professional modelers, greatly reducing labor costs, while saving a lot of time and improving work efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1a A schematic diagram of assigning different colors to different tissues in a heart model provided in an embodiment of this specification;

[0044] Figure 1b A flowchart of a method for generating texture of a medical image model provided in an embodiment of this specification;

[0045] Figure 2 A schematic diagram of a process for generating texture of a medical image model provided in an embodiment of this specification;

[0046] Figure 3a A schematic diagram of a process for determining a mapping three-dimensional grid model provided in an embodiment of this specification;

[0047] Figure 3bA schematic diagram of a portion of images in a cardiac medical image sequence provided in an embodiment of this specification;

[0048] Figure 3c A schematic diagram of a rendering model including color texture information corresponding to the entire heart provided in an embodiment of this specification;

[0049] Figure 3d A schematic diagram of a portion of an image in a rendered image sequence provided in an embodiment of this specification;

[0050] Figure 4a A schematic diagram of a process for generating a mapped three-dimensional mesh model provided in an embodiment of this specification;

[0051] Figure 4b A schematic diagram of a three-dimensional grid model provided for an embodiment of this specification;

[0052] Figure 5a A schematic diagram of a process for generating a color map provided in an embodiment of this specification;

[0053] Figure 5b A schematic diagram of an initial three-dimensional mesh model provided for an embodiment of this specification;

[0054] Figure 5c A schematic diagram of a registered three-dimensional mesh model provided in an embodiment of this specification;

[0055] Figure 6a A schematic diagram of a process for generating a color map provided in an embodiment of this specification;

[0056] Figure 6b A schematic diagram of a color map of the myocardium provided in an embodiment of the present specification;

[0057] Figure 6c A schematic diagram of a color map of the left ventricle provided for an embodiment of the present specification;

[0058] Figure 7a A schematic diagram of a process for generating a target three-dimensional mesh model with concave-convex texture information provided in an embodiment of this specification;

[0059] Figure 7b A schematic diagram of a target three-dimensional mesh model with concave-convex texture information provided in an embodiment of this specification;

[0060] Figure 8a A schematic diagram of a process for generating a normal vector texture map provided in an embodiment of this specification;

[0061] Figure 8b A schematic diagram of a normal vector texture map of the myocardium provided in an embodiment of this specification;

[0062] Figure 8c A schematic diagram of a normal vector texture map of the left ventricle provided in an embodiment of this specification;

[0063] Fig. 9 A schematic diagram of an initial three-dimensional mesh model corresponding to the entire heart provided in an embodiment of this specification;

[0064] Fig.10 A schematic diagram of an initial three-dimensional mesh model corresponding to myocardial tissue provided in an embodiment of this specification and an initial map corresponding to the myocardial tissue after it is unfolded;

[0065] Fig.11a A schematic diagram of an initial three-dimensional mesh model provided for an embodiment of this specification;

[0066] Fig.11b A schematic diagram of a rendering model including color texture information corresponding to the entire heart provided in an embodiment of this specification;

[0067] Fig.11c A schematic diagram of a target three-dimensional mesh model corresponding to the entire heart including color texture information provided in an embodiment of this specification;

[0068] Fig.12 A schematic diagram of a process for generating texture of a medical image model provided in an embodiment of this specification;

[0069] Fig.13 A schematic diagram of a texture generation device for a medical image model provided in an embodiment of this specification. DETAILED DESCRIPTION

[0070] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and should not be construed as limiting the present invention.

[0071] In medical diagnosis, although two-dimensional tomographic medical image sequences can show the relationship between pixels within the section, they cannot intuitively display the three-dimensional structure of complex organs. In order to overcome this limitation, three-dimensional reconstruction technology is widely used. Its main purpose is to convert two-dimensional images into three-dimensional models with a sense of three-dimensionality, so that doctors can observe the three-dimensional image of the lesion from multiple angles in depth, so as to analyze the condition more comprehensively and accurately. At the same time, it also helps patients and their families to more intuitively understand the patient's condition and subsequent treatment plans. However, the three-dimensional mesh model obtained by medical image reconstruction is usually grayish white and does not have texture features, which makes the model lack of realism, which is not ideal for observing different tissues and lesions in the model.

[0072] In the related technology, there are two main algorithms for three-dimensional reconstruction of medical images: one is the volume rendering algorithm, such as the ray casting algorithm and the shear-deformation algorithm, which can provide rich visual information and generate a three-dimensional model with realism and stereoscopic effect.

[0073] However, volume rendering algorithms require real-time rendering, have high computational complexity, require a lot of computing resources and time, and have relatively limited application scenarios.

[0074] The other is the surface rendering algorithm, such as the Marching Cubes algorithm, which can generate 3D models that can be exported to common 3D model formats (such as OBJ, STL, etc.), suitable for scenes such as surgical simulation and teaching demonstration. However, the 3D mesh model constructed by the surface rendering algorithm lacks texture features and realism.

[0075] In order to solve the problem that the 3D mesh model generated by the surface rendering algorithm lacks texture features, two solutions are mainly used: one is to manually or automatically assign colors to different tissues in the reconstructed model through medical image processing software (such as ITK-SNAP, 3D Slicer, etc.) to distinguish different tissues in the model. For example, please refer to Figure 1a , Figure 1a The first one shows a schematic diagram of assigning different colors to different tissues in the heart model. The second one is to use professional modeling tools to manually draw texture maps to improve the visual effect and expressiveness of the model.

[0076] However, manual or automatic color assignment methods are based on simple algorithms or completely subjective judgments, assigning the same color to the same organ tissue, which is insufficient to show the true texture details of the tissue.

[0077] The method of manually drawing texture maps requires professional modelers to manually draw them, and cannot be automatically processed in batches, so it cannot be applied to full-process automated algorithm processing. At the same time, the method of manually drawing texture maps is costly, time-consuming, and difficult to achieve ideal results when processing complex mesh models.

[0078] Based on this, the embodiment of this specification provides a texture generation method for a medical image model. First, a model is constructed for the target object to obtain an initial three-dimensional mesh model and a mapped three-dimensional mesh model containing color texture information. Then, based on multiple vertices of the initial three-dimensional mesh model, the initial three-dimensional mesh model is mapped to a two-dimensional plane to obtain an initial map corresponding to the initial three-dimensional mesh model. Then, based on the mapped three-dimensional mesh model, the initial map is assigned color to generate a color map. Finally, the color map is applied to the initial three-dimensional mesh model to obtain a target three-dimensional mesh model containing color texture information.

[0079] The above process completes the reconstruction, mapping and expansion, color map generation and other tasks of the three-dimensional mesh model through automated processing, thereby improving processing efficiency and reducing the need for manual intervention. By applying the color map to the initial three-dimensional mesh model, the target three-dimensional mesh model containing color texture information is similar to the actual result, has a high sense of reality, and accurately reflects the texture characteristics of the tissue, thereby improving the visual effect and expressiveness of the three-dimensional mesh model, providing a more accurate and intuitive reference for clinical diagnosis and surgical planning, and helping to improve the diagnosis and treatment effects. Compared with the method of manually drawing texture maps, this embodiment does not require the participation of professional modelers, greatly reducing labor costs, while saving a lot of time and improving work efficiency.

[0080] For example, see Figure 1b First, a medical image sequence is received. The input medical image sequence is modeled using a surface rendering method to obtain a mesh model without color texture, and the mesh model is saved in .obj format. The UV coordinates of the model are then expanded so that the storage information of each vertex has corresponding UV coordinates in addition to geometric coordinates, which is convenient for subsequent texture mapping, and a mesh model A is obtained for standby.

[0081] At the same time, the input medical image sequence is modeled by volume rendering method to obtain a rendering model with color texture, and then the volume rendering result is captured and saved around 360° using batch saving technology. Then, the above screenshots are reconstructed in three dimensions using the three-dimensional object reconstruction method based on multiple views to obtain a three-dimensional mesh model with color texture mapping that is similar to the original volume rendering model, which is also saved as a mesh model B in .obj format.

[0082] Next, the processed face-drawing mesh model A is registered and aligned with the mesh model B based on multi-view reconstruction using the registration method. Then, each vertex in mesh model A is traversed, and the nearest neighbor search algorithm is used to find the nearest vertex on model B, and the color of the vertex of model B is copied. Then, the color of each vertex in model A is assigned using the color of the vertex of model B, and the color texture map of model A can be generated based on this. The color map is applied to the mesh model without color texture to obtain the target 3D mesh model containing color texture information.

[0083] Finally, in order to enhance the realism of the model, the normal vector texture map of model A is generated based on image processing technology to simulate the concave and convex feeling of the surface. The normal vector texture map is applied to the target 3D mesh model so that the target 3D mesh model has concave and convex texture information.

[0084] This specification provides a method for generating texture of a medical image model. Figure 2, the texture generation method of the medical image model may include the following steps:

[0085] S210, obtaining an initial three-dimensional mesh model and a mapping three-dimensional mesh model including color texture information.

[0086] The initial 3D mesh model and the textured 3D mesh model are both obtained by modeling the target object, which may be a tissue structure, such as a heart.

[0087] Specifically, the target object is scanned by a medical imaging device, and then a three-dimensional modeling method is used to model the image data obtained by the scan to obtain an initial three-dimensional mesh model. The target object is scanned by a medical imaging device, and then a multi-view three-dimensional reconstruction method is used to model the image data obtained by the scan to obtain a mapped three-dimensional mesh model containing color texture information. The initial three-dimensional mesh model obtained by model construction and the mapped three-dimensional mesh model containing color texture information can be transmitted to a server for storage, thereby obtaining the initial three-dimensional mesh model and the mapped three-dimensional mesh model containing color texture information. In other embodiments, the server locally stores several three-dimensional mesh models obtained by model construction for the target object and a three-dimensional mesh model containing color texture information, and obtains the initial three-dimensional mesh model and the mapped three-dimensional mesh model containing color texture information from the three-dimensional mesh model and the three-dimensional mesh model containing color texture information.

[0088] It should be noted that the initial 3D mesh model can be saved in .obj format, which is a common 3D model file format suitable for use in a variety of applications.

[0089] S220 , based on multiple vertices of the initial three-dimensional mesh model, mapping the initial three-dimensional mesh model to a two-dimensional plane to obtain an initial map corresponding to the initial three-dimensional mesh model.

[0090] S230: Based on the texture three-dimensional grid model, assign colors to the initial texture to generate a color texture.

[0091] S240, applying the color map to the initial three-dimensional mesh model to obtain a target three-dimensional mesh model containing color texture information.

[0092] Specifically, the initial three-dimensional mesh model contains a plurality of vertices, which will serve as the basis for mapping. Using these vertices, the initial three-dimensional mesh model is mapped to a two-dimensional plane through parameterization technology to obtain the initial texture corresponding to the initial three-dimensional mesh model. Then, by matching the vertices between the texture three-dimensional mesh model and the initial texture, the corresponding vertices are determined, and the vertices of the initial texture are assigned colors based on the color information of the vertices of the texture three-dimensional mesh model. Based on the vertex colors of the initial texture, a color map is generated. Finally, the color map is applied to the initial three-dimensional mesh model, and the texture map is combined with the initial three-dimensional mesh model to obtain the target three-dimensional mesh model containing color texture information.

[0093] In addition, the target 3D mesh model can be directly applied to medical scenarios such as naked-eye 3D display, surgical simulation, and teaching demonstration.

[0094] Exemplarily, the initial three-dimensional mesh model can be saved in .obj format. There is a .mtl file between the color map and the initial three-dimensional mesh model, which is used to store the mapping relationship between the initial three-dimensional mesh model and the color map. In actual use, as long as the color map, the initial three-dimensional mesh model and the .mtl file are ensured to exist at the same time, the software will automatically read the .mtl file associated with the initial three-dimensional mesh model and the map file specified in the .mtl file, and automatically combine the texture map with the initial three-dimensional mesh model. At this time, the .obj format file is the target three-dimensional mesh containing color texture information.

[0095] In the above implementation, a model is first constructed for the target object to obtain an initial three-dimensional mesh model and a mapping three-dimensional mesh model containing color texture information. Then, based on multiple vertices of the initial three-dimensional mesh model, the initial three-dimensional mesh model is mapped to a two-dimensional plane to obtain an initial map corresponding to the initial three-dimensional mesh model. Then, based on the mapping three-dimensional mesh model, the initial map is assigned a color value to generate a color map. Finally, the color map is applied to the initial three-dimensional mesh model to obtain a target three-dimensional mesh model containing color texture information.

[0096] The above process completes the reconstruction, mapping and expansion, color map generation and other tasks of the three-dimensional mesh model through automated processing, thereby improving processing efficiency and reducing the need for manual intervention. By applying the color map to the initial three-dimensional mesh model, the target three-dimensional mesh model containing color texture information is similar to the actual result, has a high sense of reality, and accurately reflects the texture characteristics of the tissue, thereby improving the visual effect and expressiveness of the three-dimensional mesh model, providing a more accurate and intuitive reference for clinical diagnosis and surgical planning, and helping to improve the diagnosis and treatment effects. Compared with the method of manually drawing texture maps, this embodiment does not require the participation of professional modelers, greatly reducing labor costs, while saving a lot of time and improving work efficiency.

[0097] In some embodiments, see Figure 3a , the mapping 3D mesh model is determined by:

[0098] S310: Render a medical image sequence including a target object to generate a rendering model including color and texture information.

[0099] Specifically, the target object required is determined according to the actual situation and requirements. The target object is scanned by a medical imaging device to obtain a medical image sequence including the target object. The medical image sequence including the target object can be transmitted to a server for storage, thereby obtaining a medical image sequence including the target object. In other embodiments, the server locally stores several groups of initial medical image sequences including the target object, and obtains a medical image sequence including the target object from the initial medical image sequence including the target object. The medical image sequence is rendered using a volume rendering method to generate a rendering model containing color texture information.

[0100] For example, the target object may be a heart. Take a set of 220 cardiac medical image sequences as an example. In this set of medical image sequences, the heart tissue is divided into 9 parts, namely, ascending aorta, coronary artery, myocardium, left ventricle, right ventricle, right ventricle (upper), left atrium, right atrium, and aorta. Figure 3b , Figure 3b Some images in this set of cardiac medical image sequences are shown. The cardiac medical image sequence is rendered using the ray casting method to generate a rendering model containing color texture information corresponding to each tissue in the heart. Then the rendering models containing color texture information corresponding to each tissue in the heart are combined together to obtain a rendering model containing color texture information corresponding to the entire heart. Figure 3c , Figure 3c This is a rendering model corresponding to the entire heart, which contains color texture information.

[0101] S320 , acquiring images corresponding to the rendering model from different angles to form a rendering image sequence.

[0102] S330, reconstructing the mesh based on the rendered image sequence to generate a textured three-dimensional mesh model.

[0103] Specifically, since the rendering model containing color texture information itself is not a static, storable model file. Every time observation or analysis is required, the medical image sequence including the target object needs to be re-rendered in real time to generate a rendering model containing color texture information. Therefore, after the rendering model is generated, in order to utilize the information of the rendering model, a surround interception method can be used, that is, to capture images of the rendering model from different angles, and then save the intercepted images to form a rendering image sequence. Since the rendering image sequence contains the color information of the rendering model, in order to apply its color information to the initial three-dimensional mesh model, the rendering image sequence is used to perform further mesh reconstruction operations to generate a storable textured three-dimensional mesh model with realistic color texture information.

[0104] For example, the batch save function can be used to save a complete 360° surround image of the rendered model corresponding to the entire heart, including color and texture information. For example, by intercepting the rendering results at different angles, 50 images are saved for each tissue of the heart, thereby forming a rendered image sequence. The rendered image sequence can be saved in BMP format, see Figure 3d , Figure 3d Shows some images from the rendered image sequence.

[0105] In the above embodiment, a medical image sequence including a target object is rendered to generate a rendering model including color and texture information, images corresponding to the rendering model are acquired from different angles to form a rendering image sequence, and a mesh is reconstructed based on the rendering image sequence to generate a three-dimensional mesh model with color and texture information that is close to reality, thereby providing a data basis for subsequently obtaining a target three-dimensional mesh model including color and texture information.

[0106] In some embodiments, see Figure 4a , based on the rendered image sequence, mesh reconstruction is performed to generate a textured three-dimensional mesh model, which may include the following steps:

[0107] S410 , performing motion analysis and conversion based on the rendered image sequence, and obtaining a depth image sequence corresponding to the rendered image sequence.

[0108] S420, performing dense reconstruction based on the depth image sequence to generate a textured three-dimensional mesh model.

[0109] Specifically, when generating a rendered image sequence, not all visual information in the rendering model is fully included, so it is necessary to determine their exact positions in the three-dimensional space by analyzing the relative positions of each rendered image in the rendered image sequence. By performing motion analysis on the rendered image sequence, the relative positions of each rendered image in the rendered image sequence can be determined. Then, based on the relative positions of each rendered image in the rendered image sequence, the rendered image sequence is transformed to obtain a depth image sequence corresponding to the rendered image sequence. However, the depth image sequence only contains part of the scene information and has less information data, so dense reconstruction based on the depth image sequence is required to generate a mapped three-dimensional mesh model.

[0110] For example, the Structure from Motion (SfM) algorithm is first used to analyze the geometric relationship between the rendered image sequences to restore the camera's motion trajectory and the three-dimensional structure of the scene, and then Multi-view Stereo Matching (MVS) is used to match images from different perspectives to generate a dense three-dimensional point cloud, and finally reconstruct a complete textured three-dimensional mesh model. The specific process is as follows: Use the SIFT feature extraction algorithm to extract feature points from the rendered image sequence, and then match the corresponding feature points between different rendered images in the rendered image sequence, that is, minimize the reprojection error and obtain the motion transformation matrix. The reprojection error is shown in the following formula:

[0111]

[0112] Among them, T * is the motion transformation matrix, u i is the observed position of feature point i in the rendered image, s i is the depth value of feature point i, and K is the internal parameter matrix of the camera. i is the three-dimensional coordinate of feature point i in the camera coordinate system. Based on the motion transformation matrix T * , adjacent rendered images are determined in the rendered image sequence, and then the disparity d(x, y) is determined based on the adjacent rendered images.

[0113] Each rendered image in the rendered image sequence is converted into a depth map to obtain a depth image sequence. For each pixel position (x, y) in the rendered image, its corresponding depth value Z(x, y) in the depth map is:

[0114]

[0115] Where B is the baseline length, f is the focal length, and d(x,y) is the parallax.

[0116] Multi-view Stereo Matching (MVS) is used to match images from different perspectives based on depth image sequences to generate dense 3D point clouds. A mesh reconstruction method is used to generate a textured 3D mesh model from the dense 3D point cloud. See Figure 4b , Figure 4b Shows a textured 3D mesh model.

[0117] In the above implementation, motion analysis and transformation are performed based on the rendered image sequence, a depth image sequence corresponding to the rendered image sequence is obtained, and dense reconstruction is performed based on the depth image sequence to generate a three-dimensional mesh model with realistic color and texture information, which provides a data basis for the subsequent acquisition of a target three-dimensional mesh model containing color and texture information.

[0118] In some embodiments, see Figure 5a , based on the texture three-dimensional grid model, assigning colors to the initial texture and generating a color map may include the following steps:

[0119] S510 , using the initial three-dimensional mesh model as a reference, aligning the mapped three-dimensional mesh model to obtain a registered three-dimensional mesh model.

[0120] S520: Assign color to the initial map using the registered three-dimensional mesh model to generate a color map.

[0121] Specifically, the mapped 3D mesh model has realistic color and texture information in appearance, but its overall geometric shape is relatively rough and inaccurate. In contrast, the initial 3D mesh model does not have color and texture information, but its overall geometric shape is more accurate and more realistic. Therefore, it is necessary to migrate the color and texture information on the mapped 3D mesh model to the initial 3D mesh model, so that the accuracy of the overall shape of the model can be maintained while having realistic color and texture information.

[0122] According to the specific needs and data characteristics, an appropriate registration algorithm (such as point cloud registration, feature point registration) is selected. The initial three-dimensional mesh model is used as a reference, and the selected registration algorithm is used to align the mapping three-dimensional mesh model to obtain a registered three-dimensional mesh model. The overall geometric shape of the registered three-dimensional mesh model matches that of the initial three-dimensional mesh model. In addition, since the initial map is obtained based on the initial three-dimensional mesh model, there is a mapping relationship between the initial map and the initial three-dimensional mesh model. Therefore, using the registered three-dimensional mesh model, the color of the initial three-dimensional mesh model corresponding to it can be determined, thereby assigning color to the initial map. After the initial map is assigned color as a whole, a color map is generated.

[0123] For example, the initial 3D mesh model is used as a reference, and the initial 3D mesh model and the mapped 3D mesh model are aligned and scaled to the same size. Then, the two models are registered using the iterative closest point (ICP) algorithm, and the distance between points is kept within a certain threshold range through continuous iterations, where the registration criterion is based on the least squares method. Finally, the scaled mapped 3D mesh model is rotated and translated based on the registration results of the points between the models to obtain the registered 3D mesh model. Figure 5b , Figure 5b shows the initial 3D mesh model, see Figure 5c , Figure 5c Shown with Figure 5b The initial 3D mesh model is used as the registered 3D mesh model of the benchmark.

[0124] The registration principle of the iterative closest point (ICP) algorithm is as follows: Assume that there are two point cloud models P and Q to be registered. First, randomly select a point in the point cloud P, denoted as pi. Then find a point with the shortest Euclidean distance from pi from the point cloud Q, denoted as qi. Pi and qi are regarded as corresponding points, and the rotation transformation matrix and translation transformation matrix are obtained based on all corresponding point pairs pi and qi. After multiple iterations, the optimal rotation transformation matrix and translation transformation matrix are finally obtained, so that the point cloud models P and Q coincide. Among them, the formula for the iteration termination condition is as follows:

[0125]

[0126] Among them, R represents the rotation transformation matrix, T represents the translation transformation matrix, and k represents the number of points in the point cloud model.

[0127] In the above implementation, the initial 3D mesh model is used as a reference to facilitate the final combination operation. Then the texture 3D mesh model is aligned to obtain a registered 3D mesh model, and the registered 3D mesh model is used to assign colors to the initial texture to generate a color map with realistic color texture information, which is a data basis for the subsequent target 3D mesh model containing color texture information.

[0128] In some embodiments, see Figure 6a , using the registered three-dimensional mesh model to assign colors to the initial map and generate a color map, the following steps may be included:

[0129] S610: Determine, on the registered three-dimensional mesh model, registration vertices corresponding to vertices of the initial three-dimensional mesh model.

[0130] S620: Assign colors to corresponding vertices of the initial three-dimensional mesh model using the registration vertices.

[0131] S630: Based on the mapping relationship between the initial three-dimensional mesh model and the initial map, the initial map is assigned colors using the vertices of the initial three-dimensional mesh model after color assignment to generate a color map.

[0132] Specifically, a vertex is selected from the initial three-dimensional mesh model, and the vertex closest to the vertex is determined in the registered three-dimensional mesh model by a distance calculation method as the registration vertex corresponding to the vertex of the initial three-dimensional mesh model. All vertices of the initial three-dimensional mesh model are repeatedly traversed, and the above registration operation is performed. The registration vertex corresponding to each vertex of the initial three-dimensional mesh model can be determined on the registered three-dimensional mesh model. Subsequently, the color information of the registration vertex is used to assign values ​​to the corresponding vertices on the initial three-dimensional mesh model. Repeat the above operation to assign colors to all vertices of the initial three-dimensional mesh model.

[0133] The initial map is obtained based on the initial 3D mesh model, so there is a mapping relationship between the initial 3D mesh model and the initial map. The mapping relationship indicates that the vertices of the initial 3D mesh model have corresponding positions on the initial map. Using the mapping relationship between the initial 3D mesh model and the initial map, the colors of the vertices on the initial 3D mesh model are assigned to the corresponding positions on the initial map. After the colors of all vertices on the initial 3D mesh model are assigned to the corresponding positions on the initial map, the remaining positions without color assignment are processed using interpolation and other methods to generate a final color map with realistic color texture information.

[0134] Exemplarily, for vertex Va on the initial 3D mesh model, find its nearest vertex Vb on the registered 3D mesh model. Then assign the color value of vertex Vb on the registered 3D mesh model to vertex Va on the initial 3D mesh model. Repeat the traversal of all vertices of the initial 3D mesh model and perform the above color assignment operation to assign colors to all vertices of the initial 3D mesh model. Then, based on the mapping relationship between the vertex coordinates (x, y, z) on the initial 3D mesh model and the coordinates (U, V) of the vertices on the initial map, use the vertices of the initial 3D mesh model after color assignment to assign colors to the initial map to generate a color map. Please refer to Figure 6b , Figure 6b A color map showing the myocardium is shown in Figure 6c , Figure 6c The color map of the left ventricle is shown. The distance calculation formula is as follows:

[0135]

[0136] Among them, p 1x Represents the x-axis coordinate of the vertex on the initial 3D mesh model, p 1yRepresents the y-axis coordinate of the vertex on the initial 3D mesh model, p 1z Represents the z-axis coordinate of the vertex on the initial three-dimensional mesh model, p 2x represents the x-axis coordinate of the vertex on the 3D mesh model after registration, p 2y represents the y-axis coordinate of the vertex on the 3D mesh model after registration, p 2z Represents the z-axis coordinate of the vertex on the 3D mesh model after registration.

[0137] In the above implementation, on the registered three-dimensional mesh model, the registration vertices corresponding to the vertices of the initial three-dimensional mesh model are determined, and the corresponding vertices of the initial three-dimensional mesh model are assigned colors using the registration vertices. Based on the mapping relationship between the initial three-dimensional mesh model and the initial map, the vertices of the initial three-dimensional mesh model after color assignment are used to assign colors to the initial map, and a color map with color texture information that tends to be realistic is generated, which provides a data basis for subsequently obtaining a target three-dimensional mesh model containing color texture information.

[0138] In some embodiments, see Figure 7a , the method may include the following steps:

[0139] S710: Generate a normal vector texture map corresponding to the initial three-dimensional mesh model.

[0140] S720 , applying the normal vector texture map to the target three-dimensional mesh model so that the target three-dimensional mesh model has concave-convex texture information.

[0141] Specifically, in order to enhance the surface concavity and convexity of the target 3D mesh model and enhance the overall texture of the target 3D mesh model, it is necessary to process the color map and generate a normal vector texture map corresponding to the initial 3D mesh model. The normal vector texture map is applied to the target 3D mesh model and combined with the target 3D mesh model so that the target 3D mesh model has concavity and convexity texture information. The target 3D mesh model with concavity and convexity texture information can be used to adjust the lighting and other information in detail. For example, please refer to Figure 7b , Figure 7b It is a target 3D mesh model with bump texture information.

[0142] In the above implementation, a normal vector texture map corresponding to the initial three-dimensional mesh model is generated, and the normal vector texture map is applied to the target three-dimensional mesh model, so that the target three-dimensional mesh model has concave and convex texture information, thereby improving the visual effect and expressiveness of the three-dimensional mesh model, providing a more accurate and intuitive reference for clinical diagnosis and surgical planning, and helping to improve diagnostic and treatment effects.

[0143] In some embodiments, see Figure 8a, generating a normal vector texture map corresponding to the initial three-dimensional mesh model, can include the following steps:

[0144] S810, convert the color map into a grayscale map.

[0145] S820: Generate a normal vector texture map based on the gradient information of the grayscale map.

[0146] Specifically, the color map is usually an image in RGB format, and each pixel has color information of three channels: red, green, and blue. For each pixel of the color map, the RGB color is converted to grayscale value using a conversion formula to obtain a grayscale map. Then the horizontal and vertical gradients of each pixel in the grayscale map are calculated, and the normal vector texture map is generated based on this. For example, see Figure 8b , Figure 8b Shows the normal texture map of the myocardium, see Figure 8c , Figure 8c The normal vector texture map of the left ventricle is shown. It should be noted that since all normal lines point toward the z-axis (0, 0, 1), the generated normal vector texture maps are all blue.

[0147] In the above implementation, the color map is converted into a grayscale map, and a normal vector texture map is generated based on the gradient information of the grayscale map to enhance the detailed information of the texture.

[0148] In some embodiments, the initial three-dimensional mesh model may be determined by performing surface rendering modeling on a medical image sequence including the target object to generate the initial three-dimensional mesh model.

[0149] Specifically, the target object required is determined according to the actual situation and requirements. The target object is scanned by a medical imaging device to obtain a medical image sequence including the target object. The medical image sequence including the target object can be transmitted to a server for storage, thereby obtaining a medical image sequence including the target object. In other embodiments, the server locally stores several groups of initial medical image sequences including the target object, and obtains a medical image sequence including the target object from the initial medical image sequence including the target object. The medical image sequence is subjected to surface rendering modeling using a surface rendering method to generate an initial three-dimensional mesh model.

[0150] For example, the MC reconstruction algorithm (Marching Cubes) is used to perform surface rendering modeling on a cardiac medical image sequence to generate an initial 3D mesh model corresponding to each tissue in the heart. The initial 3D mesh model corresponding to each tissue in the heart is stored in .obj format and arranged according to the corresponding position coordinates in the image. Then, the initial 3D mesh models corresponding to each tissue in the heart are combined together and automatically arranged according to the position of each tissue to obtain an initial 3D mesh model corresponding to the entire heart. The initial 3D mesh model is stored in .obj format. Please refer to Fig. 9 , Fig. 9 The initial three-dimensional mesh model corresponding to the entire heart is shown.

[0151] In the above implementation, surface rendering modeling is performed on a medical image sequence including a target object to generate an initial three-dimensional mesh model, which provides a basis for the subsequent generation of a target three-dimensional mesh model.

[0152] In some embodiments, based on multiple vertices of the initial three-dimensional mesh model, the initial three-dimensional mesh model is mapped to a two-dimensional plane to obtain an initial map corresponding to the initial three-dimensional mesh model, which may include: based on multiple vertices of the initial three-dimensional mesh model, the three-dimensional model surface is divided into multiple fragments, and each fragment is mapped to a two-dimensional plane to obtain an initial map.

[0153] Specifically, the initial three-dimensional mesh model contains multiple vertices. The surface of the initial three-dimensional mesh model is divided into multiple fragments, each fragment contains triangles of similar shape and size, and the number of boundaries and the length of boundaries between the fragments are minimized to ensure that the divided fragments have similar shapes and sizes. Each fragment is mapped to a two-dimensional plane through a parameterization technique. Deformation and overlap are minimized to obtain an initial map. The multiple vertices contained in the initial three-dimensional mesh model have corresponding positions in the initial map, forming a mapping relationship.

[0154] Exemplarily, the specific unfolding process of mapping the vertices (X, Y, Z) on the initial three-dimensional mesh model to the (U, V) coordinates on the two-dimensional map using the UVatlas automatic UV unfolding algorithm is as follows: First, the surface of the initial three-dimensional mesh model is divided into multiple fragments, each fragment contains triangles of similar shape and size, and the number and length of boundaries between fragments are minimized to ensure that the divided fragments have similar shapes and sizes, thereby laying the foundation for subsequent parametric mapping. Then, the LSCM (Least Squares Conformal Mapping) parametric method is used to map each fragment to a two-dimensional plane. The goal of the LSCM method is to minimize deformation energy. The formula of the LSCM method is as follows:

[0155]

[0156] where z i Represents the vertex coordinates of the initial 3D mesh model, z i1 Represents the coordinates of the parameterized fragment on the two-dimensional plane, and R is a rotation matrix used to transform the vertex coordinates z of the initial three-dimensional mesh model i The parameterized coordinate z mapped to the two-dimensional plane i1 .

[0157] The parameterized fragments are unfolded onto a two-dimensional plane to minimize deformation and overlap. The unfolding process can be achieved by optimizing an energy function, as shown below:

[0158]

[0159] Among them, θ ij Represents the angle between the edges connecting vertices i and j on the unfolded plane.

[0160] See also Fig.10 , Fig.10 The initial three-dimensional mesh model corresponding to the myocardial tissue and the initial map corresponding to the unfolded myocardial tissue are shown.

[0161] In the above implementation, based on multiple vertices of the initial three-dimensional mesh model, the surface of the three-dimensional model is divided into multiple fragments, and each fragment is mapped to a two-dimensional plane to obtain an initial map, providing a basis for subsequently obtaining a target three-dimensional mesh model containing color texture information.

[0162] In another example, a heart is scanned by a medical imaging device to obtain a sequence of cardiac medical images. The cardiac medical image sequence is modeled by surface rendering using a surface rendering method to generate an initial three-dimensional mesh model corresponding to each tissue in the heart. The initial three-dimensional mesh model corresponding to each tissue in the heart is stored in the .obj format and arranged according to the corresponding position coordinates in the image. Then, the initial three-dimensional mesh models corresponding to each tissue in the heart are combined together and automatically arranged according to the position of each tissue to obtain an initial three-dimensional mesh model corresponding to the entire heart. The initial three-dimensional mesh model is stored in the .obj format. Please refer to Fig.11a , Fig.11a The initial 3D mesh model is shown. Based on the multiple vertices of the initial 3D mesh model, the 3D model surface is divided into multiple fragments, and the UVatlas automatic UV unfolding algorithm is used to map the vertices (X, Y, Z) on the initial 3D mesh model to the (U, V) coordinates on the 2D map, so that each fragment is mapped to the 2D plane to obtain the initial map.

[0163] The natural volume rendering method is used to render the cardiac medical image sequence to generate a rendering model containing color and texture information corresponding to each tissue in the heart. Then the rendering models containing color and texture information corresponding to each tissue in the heart are combined together to obtain a rendering model containing color and texture information corresponding to the entire heart. Fig.11b , Fig.11b The rendering model corresponding to the whole heart containing color texture information is displayed. Then, images of the rendering model are captured from different angles, and the captured images are saved to form a rendering image sequence. By performing motion analysis on the rendering image sequence, the relative position between each rendering image in the rendering image sequence can be determined. Then, based on the relative position between each rendering image in the rendering image sequence, the rendering image sequence is transformed to obtain a depth image sequence corresponding to the rendering image sequence. However, the depth image sequence only contains part of the scene information and the information data is less, so dense reconstruction based on the depth image sequence is required to generate a three-dimensional mesh model of a map containing color texture information.

[0164] Then, taking the initial three-dimensional mesh model as a reference, the mapping three-dimensional mesh model is aligned and registered using a registration algorithm to obtain a registered three-dimensional mesh model. On the registered three-dimensional mesh model, the registration vertices corresponding to the vertices of the initial three-dimensional mesh model are determined. Subsequently, the corresponding vertices of the initial three-dimensional mesh model are assigned colors using the registration vertices. Based on the mapping relationship between the initial three-dimensional mesh model and the initial map, the vertices of the initial three-dimensional mesh model after color assignment are used to assign colors to the initial map to generate a color map. In addition, the color map is converted into a grayscale map. Based on the gradient information of the grayscale map, a normal vector texture map is generated.

[0165] Finally, the color map is applied to the initial 3D mesh model to obtain the target 3D mesh model containing color texture information. The normal vector texture map is applied to the target 3D mesh model so that the target 3D mesh model has concave and convex texture information. Fig.11c , Fig.11c Displays the target 3D mesh model corresponding to the entire heart with concave-convex texture information.

[0166] The present specification also provides a method for generating texture of a medical image model. For example, see Fig.12 , the texture generation method of the medical image model may include the following steps:

[0167] S1202: Perform surface rendering modeling on a medical image sequence including a target object to generate an initial three-dimensional mesh model.

[0168] Among them, the initial three-dimensional mesh model is obtained by modeling the target object.

[0169] S1204: Render the medical image sequence including the target object to generate a rendering model including color and texture information.

[0170] S1206: Acquire images corresponding to the rendering model from different angles to form a rendering image sequence.

[0171] S1208 . Perform motion analysis and conversion based on the rendered image sequence to obtain a depth image sequence corresponding to the rendered image sequence.

[0172] S1210, performing dense reconstruction based on the depth image sequence to generate a textured three-dimensional mesh model containing color texture information.

[0173] Among them, the mapped three-dimensional mesh model is obtained by modeling the target object.

[0174] S1212. Based on multiple vertices of the initial three-dimensional mesh model, the surface of the three-dimensional model is divided into multiple segments, and each segment is mapped onto a two-dimensional plane to obtain an initial map corresponding to the initial three-dimensional mesh model.

[0175] S1214: Using the initial three-dimensional mesh model as a reference, aligning the mapped three-dimensional mesh model to obtain an aligned three-dimensional mesh model.

[0176] S1216. Determine, on the registered three-dimensional mesh model, registration vertices corresponding to vertices of the initial three-dimensional mesh model.

[0177] S1218. Use the registration vertices to assign colors to corresponding vertices of the initial three-dimensional mesh model.

[0178] S1220 , based on the mapping relationship between the initial three-dimensional mesh model and the initial map, use the vertices of the initial three-dimensional mesh model after color assignment to assign colors to the initial map to generate a color map.

[0179] S1222. Apply the color map to the initial three-dimensional mesh model to obtain a target three-dimensional mesh model containing color texture information.

[0180] S1224. Convert the color map into a grayscale map.

[0181] S1226. Generate a normal vector texture map based on the gradient information of the grayscale map.

[0182] S1228. Apply the normal vector texture map to the target three-dimensional mesh model so that the target three-dimensional mesh model has concave-convex texture information.

[0183] This specification embodiment provides a texture generation device 1300 for a medical image model. Fig.13The texture generation device 1300 of the medical image model includes: a mesh model acquisition module 1310, an initial map generation module 1320, a color map generation module 1330, and a target mesh model generation module 1340.

[0184] A mesh model acquisition module 1310 is used to acquire an initial 3D mesh model and a textured 3D mesh model including color and texture information; wherein both the initial 3D mesh model and the textured 3D mesh model are obtained by modeling a target object;

[0185] An initial texture generation module 1320 is used to map the initial three-dimensional mesh model to a two-dimensional plane based on multiple vertices of the initial three-dimensional mesh model to obtain an initial texture corresponding to the initial three-dimensional mesh model;

[0186] A color map generation module 1330 is used to assign colors to the initial map based on the map three-dimensional grid model to generate a color map;

[0187] The target mesh model generating module 1340 is used to apply the color map to the initial three-dimensional mesh model to obtain a target three-dimensional mesh model containing color texture information.

[0188] For a detailed description of the texture generation device for a medical image model, please refer to the description of the texture generation method for a medical image model above, which will not be repeated here.

[0189] In some embodiments, a medical imaging device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the method steps in the above embodiments when executing the computer program.

[0190] An embodiment of the present specification provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of the method in any of the above embodiments are implemented.

[0191] One embodiment of the present specification provides a computer program product, which includes instructions. When the instructions are executed by a processor of a computer device, the computer device can perform the steps of the method of any of the above embodiments.

[0192] It should be noted that the logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in combination with these instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in combination with these instruction execution systems, devices or apparatuses. More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection portion with one or more wirings (electronic device), a portable computer disk box (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium and then editing, interpreting or otherwise processing in a suitable manner if necessary, and then stored in a computer memory.

Claims

1. A texture generation method for a medical image model, characterized in that: The method comprises: Acquire an initial three-dimensional mesh model and a mapped three-dimensional mesh model containing color and texture information; wherein the initial three-dimensional mesh model and the mapped three-dimensional mesh model are both obtained by model construction for the target object; Based on multiple vertices of the initial three-dimensional mesh model, mapping the initial three-dimensional mesh model to a two-dimensional plane to obtain an initial map corresponding to the initial three-dimensional mesh model; Based on the texture three-dimensional grid model, assigning colors to the initial texture to generate a color texture; The color map is applied to the initial three-dimensional mesh model to obtain a target three-dimensional mesh model containing color texture information.

2. The method according to claim 1, characterized in that The three-dimensional mesh model of the map is determined by: Rendering a sequence of medical images including the target object to generate a rendering model including color texture information; Acquire images corresponding to the rendering model from different angles to form a rendering image sequence; Mesh reconstruction is performed based on the rendered image sequence to generate the mapped three-dimensional mesh model.

3. The method according to claim 2, characterized in that The step of performing mesh reconstruction based on the rendered image sequence to generate the textured three-dimensional mesh model comprises: Performing motion analysis and conversion based on the rendered image sequence to obtain a depth image sequence corresponding to the rendered image sequence; Dense reconstruction is performed based on the depth image sequence to generate the mapped three-dimensional mesh model.

4. The method according to claim 1, characterized in that: The step of assigning color to the initial map based on the map three-dimensional grid model to generate a color map includes: Using the initial three-dimensional mesh model as a reference, aligning the mapped three-dimensional mesh model to obtain a registered three-dimensional mesh model; The registered three-dimensional mesh model is used to assign colors to the initial map to generate the color map.

5. The method according to claim 4, characterized in that The step of assigning color to the initial map by using the registered three-dimensional mesh model to generate the color map comprises: On the registered three-dimensional mesh model, determining registration vertices corresponding to vertices of the initial three-dimensional mesh model; Using the registration vertices to assign colors to corresponding vertices of the initial three-dimensional mesh model; Based on the mapping relationship between the initial three-dimensional mesh model and the initial map, the initial map is assigned colors using the vertices of the initial three-dimensional mesh model after color assignment to generate the color map.

6. The method according to claim 1, characterized in that The method further comprises: Generating a normal vector texture map corresponding to the initial three-dimensional mesh model; The normal vector texture map is applied to the target three-dimensional mesh model so that the target three-dimensional mesh model has concave-convex texture information.

7. The method according to claim 6, characterized in that The generating a normal vector texture map corresponding to the initial three-dimensional mesh model includes: Converting the color map to a grayscale map; The normal vector texture map is generated based on the gradient information of the grayscale map.

8. The method according to claim 1, characterized in that The initial three-dimensional mesh model is determined in the following manner: performing surface rendering modeling on a medical image sequence including the target object to generate an initial three-dimensional mesh model.

9. The method according to claim 1, characterized in that: The step of mapping the initial three-dimensional mesh model to a two-dimensional plane based on a plurality of vertices of the initial three-dimensional mesh model to obtain an initial map corresponding to the initial three-dimensional mesh model includes: Based on a plurality of vertices of the initial three-dimensional mesh model, the surface of the three-dimensional model is divided into a plurality of segments, and each segment is mapped onto a two-dimensional plane to obtain the initial map.

10. A texture generation device for a medical image model, characterized in that: The device comprises: A mesh model acquisition module, used to acquire an initial three-dimensional mesh model and a textured three-dimensional mesh model containing color and texture information; wherein the initial three-dimensional mesh model and the textured three-dimensional mesh model are both obtained by model construction for a target object; An initial map generation module, used for mapping the initial three-dimensional mesh model to a two-dimensional plane based on a plurality of vertices of the initial three-dimensional mesh model to obtain an initial map corresponding to the initial three-dimensional mesh model; A color map generation module, used to assign colors to the initial map based on the map three-dimensional grid model to generate a color map; The target mesh model generation module is used to apply the color map to the initial three-dimensional mesh model to obtain a target three-dimensional mesh model containing color texture information.

11. A medical imaging device, comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 9 are implemented.

12. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 9 are implemented.

Images