Automatic segmentation and expansion method for three-dimensional portrait UV chartlet

By employing an automatic segmentation and unfolding method for 3D human portrait UV mapping, and utilizing deep learning algorithms for ordered segmentation, unfolding, and image optimization, the problem of disordered and fragmented UV mapping is solved, thereby improving mapping quality and processing effects.

CN121883775APending Publication Date: 2026-04-17XIAMEN SUXIANG TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202311679725.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2023-12-08
Publication Date
2026-04-17

AI Technical Summary

Technical Problem

When existing 3D human portrait UV maps are automatically generated by existing technology programs, they are disordered and fragmented, making it difficult to perform subsequent map optimization processing. Traditional segmentation and unfolding require manual operation and the results are not good.

Method used

An automatic segmentation and unfolding method using 3D human image UV mapping is adopted, which includes steps such as data acquisition, preliminary optimization processing, evaluation, cutting, deep learning algorithm material classification, key area identification and rebaking. The method uses deep learning algorithm to perform ordered segmentation and unfolding, and performs image optimization processing before segmentation.

Benefits of technology

It achieves orderly segmentation and unfolding of 3D human portrait UV maps, improves map quality, stabilizes and enhances subsequent processing effects, and reduces reliance on artistic skills and software operation.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121883775A_ABST
    Figure CN121883775A_ABST
Patent Text Reader

Abstract

The invention discloses an automatic segmentation and expansion method for a three-dimensional portrait UV map, and belongs to the technical field of map segmentation, and the method comprises the following steps: obtaining data, carrying out the preliminary optimization processing of the map, carrying out the preliminary evaluation of the optimized map, and carrying out the certain cutting processing of the map. Learning and classifying materials by using a deep learning algorithm, then selecting and cutting required parts in the chartlet, and identifying key areas in the chartlet; according to the method, a deep learning method is adopted, ordered segmentation and expansion of the UV chartlet can be carried out according to actual business requirements, the problems that face information of three-dimensional portrait data is independently enhanced and different parts are independently processed are effectively solved, the method carries out optimization processing of the previous step on the image before chartlet segmentation, and the processing efficiency is improved. Therefore, the mapping quality can be greatly improved, the subsequent mapping processing effect is effectively and stably improved, and the overall method is good in application effect.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the field of texture segmentation technology, and particularly relates to a method for automatic segmentation and unfolding of 3D human portrait UV textures. Background Technology

[0002] UV mapping is a common image processing technique. In UV mapping, UV stands for UV texture mapping coordinates, which are similar to the X, Y, and Z axes of a spatial model. It defines the position information of each point on an image. UV mapping precisely maps each point on an image to the surface of a model object. Software then performs smooth interpolation at the gaps between points.

[0003] While current UV image processing methods can process textures, 3D portrait UV textures are automatically generated by existing technology programs, resulting in disordered and fragmented textures that are difficult to optimize later. Traditional 3D portrait segmentation and unfolding is done manually, requiring high artistic skills and complex software operation, and the application effect is not good. There is an urgent need for an automatic segmentation and unfolding method for 3D portrait UV textures. Summary of the Invention

[0004] The purpose of this invention is to address the problems that while current UV image processing methods can process textures, the 3D portrait UV textures are automatically generated by existing technology programs, resulting in disordered and fragmented textures that are difficult to optimize later. Traditional 3D portrait segmentation and unfolding is done manually, requiring high artistic skills and complex software operations, and the application effect is not good. Therefore, this invention proposes an automatic segmentation and unfolding method for 3D portrait UV textures.

[0005] To achieve the above objectives, the present invention employs the following technical solution: a method for automatic segmentation and unfolding of 3D human portrait UV mapping, comprising the following steps:

[0006] S1. Acquire the data;

[0007] S2. Perform preliminary optimization processing on the texture;

[0008] S3. Conduct a preliminary evaluation of the optimized textures;

[0009] S4. Perform certain cutting processing on the texture;

[0010] S5. Utilize deep learning algorithms to achieve material learning and classification;

[0011] S6. Then select the desired part of the cut texture;

[0012] S7. Identify key areas in the texture;

[0013] S8. Re-bake the texture;

[0014] S9. After baking, perform a final evaluation of the split and unfolded textures.

[0015] As a further description of the above technical solution:

[0016] In step S1, data is acquired, including texture information and coordinate information.

[0017] As a further description of the above technical solution:

[0018] In step S2, the texture is initially optimized. The specific steps are as follows: import the texture into the graphics optimization library, optimize the contrast of the texture, optimize the pixels of the texture, perform noise reduction on the texture, and finally perform filtering on the texture.

[0019] As a further description of the above technical solution:

[0020] In step S3, a preliminary evaluation is performed on the optimized texture, and the evaluation results are divided into excellent, good, and poor.

[0021] As a further description of the above technical solution:

[0022] In step S4, the texture is cut into pieces. The specific steps are as follows: first, the size of the texture is calculated, then the number of small texture pieces is calculated, then the holes are identified, the small texture pieces are numbered, and the long limb parts in the texture are marked.

[0023] As a further description of the above technical solution:

[0024] In step S5, deep learning algorithms are used to learn and classify materials. The specific steps are as follows: import the neural network algorithm, detect and learn the parts of the three-body model, and make labels.

[0025] As a further description of the above technical solution:

[0026] In step S6, the desired parts of the cut texture are selected, and the desired parts represent the three-body model parts learned and detected in step S5.

[0027] As a further description of the above technical solution:

[0028] In step S7, key areas in the texture are identified, including faces and areas with high density of text and image information.

[0029] As a further description of the above technical solution:

[0030] In step S8, the texture is automatically re-baked.

[0031] As a further description of the above technical solution:

[0032] In step S9, after baking, the texture after segmentation and unfolding is finally evaluated. If the final evaluation result passes the evaluation criteria, the work is completed. If the evaluation result fails the evaluation criteria, the process returns to step S4.

[0033] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0034] In this invention, the method employs deep learning to perform ordered segmentation and unfolding of UV mapping according to actual business needs. This effectively solves the problems of separately enhancing facial information and processing different parts of 3D human portrait data. Furthermore, the method performs a previous step of image optimization before mapping segmentation, which can greatly improve mapping quality and effectively and stably improve the subsequent mapping processing effect. Overall, the method has good application results. Attached Figure Description

[0035] Figure 1 A flowchart of a method for automatically segmenting and unfolding UV maps of 3D human figures. Detailed Implementation

[0036] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0037] Example 1

[0038] Please see Figure 1 This invention provides a technical solution: a method for automatic segmentation and unfolding of 3D human portrait UV mapping, comprising the following steps:

[0039] S1. Acquire data, including texture materials, application software type, and application network data;

[0040] S2. Perform preliminary optimization on the texture. The specific steps are as follows: import the texture into the graphics optimization library, optimize the contrast of the texture, optimize the pixels of the texture, perform noise reduction on the texture, and finally perform filtering on the texture.

[0041] S3. Conduct a preliminary evaluation of the optimized textures, and the evaluation results are divided into excellent, good, and poor.

[0042] S4. Perform certain cutting processing on the texture. The specific steps are as follows: first calculate the size of the texture, calculate the number of small texture pieces to be cut, then calculate the identification holes, number the small texture pieces, and mark the long limb parts in the texture.

[0043] S5. Using deep learning algorithms to learn and classify materials, the specific steps are as follows: import the neural network algorithm, detect and learn the parts of the three-body model, and make labels;

[0044] S6. Next, select the desired part in the cut texture. The desired part represents the three-body model part learned and detected in S5.

[0045] S7. Identify key areas in the texture, including faces and areas with high density of text and image information;

[0046] S8. Re-baking the texture. The software used for texture baking is TGS Texture Manager.

[0047] S9. After baking, perform a final evaluation on the split and unfolded texture. If the final evaluation result passes the evaluation criteria, the work is completed. If the evaluation result fails the evaluation criteria, return to S4.

[0048] In this embodiment, the method employs deep learning, which can perform ordered segmentation and unfolding of UV mapping according to actual business needs, effectively solving the problems of separately enhancing facial information and processing different parts of 3D human portrait data.

[0049] Example 2

[0050] Please see Figure 1 This invention provides a technical solution: a method for automatic segmentation and unfolding of 3D human portrait UV mapping, comprising the following steps:

[0051] S1. Acquire data, including texture materials, application software type, and application network data;

[0052] S2. Perform preliminary optimization on the texture. The specific steps are as follows: import the texture into the graphics optimization library, optimize the contrast of the texture, optimize the pixels of the texture, perform noise reduction on the texture, and finally perform filtering on the texture.

[0053] S3. Conduct a preliminary evaluation of the optimized textures, and the evaluation results are divided into excellent, good, and poor.

[0054] S4. Perform certain cutting processing on the texture. The specific steps are as follows: first calculate the size of the texture, calculate the number of small texture pieces to be cut, then calculate the identification holes, number the small texture pieces, and mark the long limb parts in the texture.

[0055] S5. Using deep learning algorithms to learn and classify materials, the specific steps are as follows: import the neural network algorithm, detect and learn the parts of the three-body model, and make labels;

[0056] S6. Next, select the desired part in the cut texture. The desired part represents the three-body model part learned and detected in S5.

[0057] S7. Identify key areas in the texture, including faces and areas with high density of text and image information;

[0058] S8. Re-baking the textures. The software used for texture baking is Substance Painter and ShaderMap Pro.

[0059] S9. After baking, perform a final evaluation on the split and unfolded texture. If the final evaluation result passes the evaluation criteria, the work is completed. If the evaluation result fails the evaluation criteria, return to S4.

[0060] In this embodiment, the method performs a previous optimization process on the image before texture segmentation, which can greatly improve texture quality and effectively and stably improve the subsequent texture processing effect. The overall application effect of the method is good.

[0061] Example 3

[0062] Please see Figure 1 This invention provides a technical solution: a method for automatic segmentation and unfolding of 3D human portrait UV mapping, comprising the following steps:

[0063] S1. Acquire data, including texture materials, application software type, and application network data;

[0064] S2. Perform preliminary optimization on the texture. The specific steps are as follows: import the texture into the graphics optimization library, optimize the contrast of the texture, optimize the pixels of the texture, perform noise reduction on the texture, and finally perform filtering on the texture.

[0065] S3. Conduct a preliminary evaluation of the optimized textures, and the evaluation results are divided into excellent, good, and poor.

[0066] S4. Perform certain cutting processing on the texture. The specific steps are as follows: first calculate the size of the texture, calculate the number of small texture pieces to be cut, then calculate the identification holes, number the small texture pieces, and mark the long limb parts in the texture.

[0067] S5. Using deep learning algorithms to learn and classify materials, the specific steps are as follows: import the neural network algorithm, detect and learn the parts of the three-body model, and make labels;

[0068] S6. Next, select the desired part in the cut texture. The desired part represents the three-body model part learned and detected in S5.

[0069] S7. Identify key areas in the texture, including faces and areas with high density of text and image information;

[0070] S8. Re-baking the textures. The software used for texture baking is Substance Painter, ShaderMap Pro, and TGS Texture Manager.

[0071] S9. After baking, perform a final evaluation on the split and unfolded texture. If the final evaluation result passes the evaluation criteria, the work is completed. If the evaluation result fails the evaluation criteria, return to S4.

[0072] In this embodiment, the method employs deep learning, which can perform ordered segmentation and unfolding of UV mapping according to actual business needs. This effectively solves the problems of separately enhancing facial information and processing different parts of 3D human portrait data. Furthermore, the method performs a previous step of image optimization before mapping segmentation, which can greatly improve mapping quality and effectively and stably improve the subsequent mapping processing effect. Overall, the method has good application results.

[0073] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for automatic segmentation and unwrapping of a three-dimensional portrait UV map, characterized in that: Includes the following steps: S1. Acquire the data; S2. Perform preliminary optimization processing on the texture; S3. Calculate and identify UV mapping for holes and long limb parts; S4. Cut the identified UV map; S5. Use deep learning neural networks to detect and label the parts of the human body model; S6. Cut the marked area; S7. Identify areas with high density of facial and text information in the texture as key areas and enlarge them proportionally. S8. Re-bake the texture; S9, Output UV data.

2. The three-dimensional portrait UV map automatic segmentation unwrapping method according to claim 1, wherein, In step S1, data is acquired, including texture information and coordinate information.

3. The three-dimensional portrait UV map automatic segmentation unwrapping method according to claim 1, wherein, In step S2, the texture is initially optimized. The specific steps are as follows: import the texture into the graphics optimization library, optimize the contrast of the texture, optimize the pixels of the texture, perform noise reduction on the texture, and finally perform filtering on the texture.

4. The method for automatic segmentation and unfolding of three-dimensional human portrait UV mapping according to claim 1, characterized in that, In step S3, a preliminary evaluation is performed on the optimized texture, and the evaluation results are divided into excellent, good, and poor.

5. The method for automatic segmentation and unfolding of three-dimensional human portrait UV mapping according to claim 1, characterized in that, In step S4, the texture is cut into pieces. The specific steps are as follows: first, the size of the texture is calculated, then the number of small texture pieces is calculated, then the holes are identified, the small texture pieces are numbered, and the long limb parts in the texture are marked.

6. The method for automatic segmentation and unfolding of three-dimensional human portrait UV mapping according to claim 1, characterized in that, In step S5, deep learning algorithms are used to learn and classify materials. The specific steps are as follows: import the neural network algorithm, detect and learn the parts of the three-body model, and make labels.

7. The method for automatic segmentation and unfolding of three-dimensional human portrait UV mapping according to claim 1, characterized in that, In step S6, the desired parts of the cut texture are selected, and the desired parts represent the three-body model parts learned and detected in step S5.

8. The method for automatic segmentation and unfolding of three-dimensional human portrait UV mapping according to claim 1, characterized in that, In step S7, key areas in the texture are identified, including faces and areas with high density of text and image information.

9. The method for automatic segmentation and unfolding of three-dimensional human portrait UV mapping according to claim 1, characterized in that, In step S8, the texture is automatically re-baked.

10. The method for automatic segmentation and unfolding of three-dimensional human portrait UV mapping according to claim 1, characterized in that, In step S9, after baking, the texture after segmentation and unfolding is finally evaluated. If the final evaluation result passes the evaluation criteria, the work is completed. If the evaluation result fails the evaluation criteria, the process returns to step S4.