3D Reconstruction View Synthesis Iterative Refinement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing 3D reconstruction methods face challenges in generating high-quality 3D models due to insufficient source images and limited viewpoints, leading to incomplete or low-quality reconstructions.
Innovation Solution
The method employs view synthesis to generate target images and viewpoints based on the spatial distribution of source viewpoints, using deep learning networks and iterative refinement to improve image quality and completeness.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If view synthesis is performed with high processing cost to improve image quality, then the quality of target images improves, but the processing time and computational resources increase
Solution Approach 1:
The patent applies partial action by performing view synthesis iteratively with varying processing costs. Initially, view synthesis is performed with standard processing cost to generate target images. Then, quality evaluation is conducted, and only regions or iterations with insufficient quality undergo re-synthesis with adjusted processing cost. This selective application of high processing cost avoids uniformly increasing computational resources while still improving image quality where needed.
Solution Approach 2:
The patent implements feedback through quality evaluation of generated target images. The evaluation results are used to determine whether to reperform view synthesis and how to adjust processing cost. This closed-loop feedback mechanism ensures that processing cost is increased only when necessary to improve quality, rather than always using maximum processing cost, thus balancing image quality improvement with processing time efficiency.
2Loss of information
If multiple target viewpoints are generated to improve 3D model completeness, then the completeness of 3D reconstruction improves, but the quantity of images and processing complexity increase
Solution Approach 1:
The patent applies local quality by evaluating the quality of 3D reconstruction in different regions and selectively generating additional target viewpoints only where needed. Instead of uniformly increasing the number of target viewpoints across the entire scene, the system identifies specific regions with insufficient reconstruction quality and generates additional viewpoints targeted at those areas. This approach improves overall completeness while avoiding the processing complexity increase that would result from uniformly increasing viewpoint quantity across all regions.
3Manufacturing precision
If iterative refinement of view synthesis is performed to improve image quality, then the quality of target images improves, but the processing time and computational resources increase
Solution Approach 1:
The patent applies partial action in iterative refinement by performing quality evaluation after each synthesis iteration and only reperforming view synthesis when quality thresholds are not met. Instead of always performing full iterative refinement with high processing cost, the system selectively applies additional synthesis iterations only to images or regions that require quality improvement. This approach maintains processing efficiency while still achieving high image quality where necessary.
Data Source
AI summary
A method and electronic apparatus for three-dimensional (3D) reconstruction of an object by using view synthesis are provided. The method includes obtaining source images of a scene including an object, generating a target viewpoint based on a spatial distribution of source viewpoints corresponding to the source images, generating a target image corresponding to the target viewpoint by performing the view synthesis, and generating a 3D model of the object by 3D reconstruction based on the source images and the target image.