Spanning Tree Stereo Matching for Efficient Disparity Calculation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current stereo matching methods require complex optimizations and large data sets, especially when propagating additional information from ground control points, which can lead to inefficiencies in calculating disparity and depth information across entire images.
Innovation Solution
The method employs a spanning tree structure to propagate additional information from reference pixels to general pixels, using edge costs and data costs based on brightness and disparity information, allowing for efficient calculation of accumulated data costs and improved stereo matching and up-sampling without the need for extensive data sets.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If complex optimization methods are used to propagate additional information from ground control points, then the quality of stereo matching may be improved, but the computational complexity and time consumption increase significantly
Solution Approach 1:
The patent segments the image into multiple regions based on depth information, processing each region separately with appropriate methods. This divides the complex global optimization problem into smaller, more manageable local problems, reducing overall computational complexity while maintaining matching quality in each region
Solution Approach 2:
The patent performs preliminary depth estimation using coarse-to-fine methods before conducting detailed stereo matching. This preliminary action provides initial guidance that simplifies subsequent optimization steps, reducing the search space and computational burden of the full matching process
2Measurement precision
If large data sets are used for all pixels of the first image, then the accuracy of disparity calculation may be improved, but the memory requirements and processing time increase
Solution Approach 1:
The patent applies different processing strategies to different pixels based on their local characteristics. Pixels in reliable depth regions use simplified methods with smaller data sets, while pixels in uncertain regions receive more intensive processing. This local differentiation maintains overall accuracy while reducing total data requirements
Solution Approach 2:
The patent enables pixels to determine their own processing needs based on local depth reliability metrics. Each pixel region autonomously selects appropriate matching strategies and data requirements, eliminating the need for uniform large-scale data processing across the entire image
3Ease of manufacture
If traditional cost aggregation methods are used without spanning tree structure, then the implementation is simpler, but the propagation of additional information from reference pixels is less efficient
Solution Approach 1:
The patent introduces a spanning tree structure as an intermediary framework that connects reference pixels to general pixels through optimal paths. This intermediary structure efficiently guides the propagation of additional information while maintaining a relatively simple implementation through standard tree traversal algorithms
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method and apparatus for generating a spanning tree, a method and apparatus for stereo matching, a method and apparatus for up-sampling, and a method and apparatus for generating a reference pixel are disclosed, in which a spanning tree may be generated by reference pixels, stereo matching or up-sampling may be performed based on the generated spanning tree, and a reference pixel may be generated based on a stereo video.