Hierarchical texture mapping method for multi-view 3D reconstruction

Abstract

The invention discloses a hierarchical multi-view three-dimensional reconstruction texture mapping method, which comprises the following steps: (1) unifying color difference according to the view angle by using a color transfer algorithm based on a front angle as a reference for an image sequence acquired under different views; (2) camera pose correction being performed on the image sequence afterunified chromatic aberration, and a projection transform-related least square method is constructed to solve the problem of camera pose, and the optimal camera pose is solved iteratively to obtain the image sequence after camera pose correction; (3) generating a target image sequence based on the principle of image similarity by taking the image sequence corrected by the camera posture calculatedin the step (2) as a parameter; (4) by constructing Markov random fields to synthesize the texture of triangular patches on the 3D model, the texture mapping being completed. The multi-view three-dimensional reconstruction texture mapping method provided by the invention can effectively improve the error problem of the multi-view three-dimensional reconstruction texture mapping, and has the advantages of good accuracy, strong sense of reality and high texture quality.

Description

technical field [0001] The invention relates to the field of three-dimensional reconstruction of multi-view in computer graphics, in particular to a hierarchical multi-view Figure three dimensional reconstruction texture mapping method. Background technique [0002] The so-called texture mapping is the process of mapping the pixels of the texture image to the pixels of the screen space. In the practical application of graphics, it is to simulate the details such as the color and material of the surface of the 3D model. On the one hand, the use of texture mapping technology can highlight the surface details of the 3D model and improve the realism of drawing, and on the other hand, it can simplify the process of 3D modeling. [0003] In recent years, the use of computer technology to model the human body is an important topic of 3D reconstruction. It has important application value in the fields of clothing customization, virtual reality, game entertainment, and film and tele...

AI Technical Summary

Problems solved by technology

[0006] 1. Due to different factors such as camera orientation and exposure conditions, there will be certain color differences in the image sequence under multiple viewing angles, which will lead to a certain degree of color block phenomenon in the surface texture of the model after the actual texture mapping;

[0007] 2. In the process of 3D reconstruction, each viewing angle corresponds to a camera pose. Due to the error of geometric data and the optical distortion of the camera, there will be certain errors in the camera pose during the actual correspondence between the image and the model. This will cause local texture misalignment when texture mapping;

[0008] 3. No matter how efficient and accurate the 3D reconstruction algorithm is, the actual recovered point cloud data will have slight deviations in its spatial position after registration and surface reconstruction. If texture mapping is performed directly, local areas with errors will blurry, distorted

[0009] 4. In the process of texture generation, due to the deviation between multiple views, there will be problems such as dislocation and gaps in the fusion and transition of texture boundaries, resulting in unsatisfactory final results

Images