3D Color Mesh Reconstruction with Vertex Color Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional three-dimensional modeling technologies face challenges in generating high-quality models from multi-view images due to brightness and color differences between images, leading to texture seams and distortions.
Innovation Solution
A method and apparatus for reconstructing a three-dimensional color mesh by receiving multi-view images and camera parameter information, constructing a texture map, correcting vertex color values, and applying these corrections to minimize distortions and eliminate texture seams.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If multi-view images are used for three-dimensional reconstruction, then realism of the result is enhanced by mapping color values, but texture seam phenomena occur due to brightness or color differences between images
Solution Approach 1:
The patent applies parameter changes by adjusting brightness and color parameters of multi-view images through histogram matching and color correction algorithms. This resolves the contradiction by modifying image parameters to achieve consistent appearance across different views, thereby maintaining both realism and texture continuity in the three-dimensional reconstruction.
Solution Approach 2:
The patent employs preliminary action by performing pre-processing operations on multi-view images before reconstruction, including radiometric calibration and color normalization. This advance preparation ensures that brightness and color differences are corrected beforehand, preventing texture seam phenomena while preserving the enhanced realism from color mapping.
2Manufacturing precision
If color values are mapped from multi-view images to three-dimensional mesh, then realism is significantly enhanced, but brightness or color differences cause texture seam phenomena
Solution Approach 1:
The patent introduces an intermediary color correction process that acts as a mediator between multi-view images and the three-dimensional mesh. This intermediary step includes computing average colors from multiple views and applying correction factors to ensure uniform texture appearance, thereby maintaining both color accuracy and texture uniformity.
Solution Approach 2:
The patent applies equipotentiality by equalizing the color and brightness levels across all multi-view images through normalization to a reference image. This creates a uniform color potential across all input images, eliminating texture seams while preserving the realistic color mapping to the three-dimensional mesh.
3Loss of information
If multiple multi-view images are processed for three-dimensional reconstruction, then comprehensive color information is obtained, but calculation time increases significantly
Solution Approach 1:
The patent applies segmentation by dividing the multi-view image processing into independent stages: color correction, feature extraction, and mesh reconstruction. This segmentation allows parallel processing of multiple images and reduces overall calculation time while maintaining complete color information from all views.
Solution Approach 2:
The patent employs preliminary action by pre-computing color correction parameters and feature descriptors from multi-view images before the actual reconstruction process. This advance preparation reduces the computational burden during reconstruction, enabling comprehensive use of color information from multiple views without excessive calculation time.
Data Source
AI summary
Disclosed is a method of reconstructing a three-dimensional color mesh and an apparatus for the same. According to an embodiment of the present disclosure, the method includes: receiving mesh information of an object, multiple multi-view images obtained by photographing the object at different positions, and camera parameter information corresponding to the multiple multi-view images; constructing a texture map with respect to the object on the basis of the received information and setting a texture patch referring to a color value of the same multi-view image; correcting a color value of a vertex included for each texture patch; and performing rendering with respect to the object by applying the corrected color value of the vertex to the texture map.