Stereo Matching System Edge Map Cost Volume Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Stereo matching systems face challenges in real-time computation of matching costs and optimization for reconstructing three-dimensional images from dual camera monoscopic images, requiring significant computational power and efficiency improvements for applications like robot eyes and autonomous vehicles.
Innovation Solution
A method involving downscaled images, edge map generation, and dynamic programming to efficiently calculate an initial disparity estimate, reducing computational complexity by selectively computing costs for edge pixels and using a downscale factor to reduce image dimensions, thereby enhancing processing speed.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If stereo matching systems compute matching costs and optimization in real time using full-resolution images, then measurement precision is improved, but device complexity and computational power requirements increase significantly
Solution Approach 1:
The patent segments the image processing task by first downsampling the images to a lower resolution, then selectively processing only edge pixels in the cost volume matrix generation. This segmentation allows the system to maintain precision for critical features (edges) while reducing overall computational complexity by ignoring non-edge regions.
Solution Approach 2:
The patent applies local quality by differentiating between edge pixels and non-edge pixels. Edge pixels receive full computational attention for cost calculation, while non-edge pixels are processed with reduced complexity. This local differentiation maintains measurement precision where needed without uniformly increasing device complexity across the entire image.
2Measurement precision
If the system processes high-resolution images to maintain depth accuracy, then measurement precision is improved, but processing speed decreases
Solution Approach 1:
The patent performs preliminary downsampling of the input images before generating the cost volume matrix. This preliminary action reduces the number of pixels that need to be processed in subsequent steps, thereby increasing processing speed while maintaining depth accuracy through the preservation of edge information in the downscaled images.
Solution Approach 2:
The patent applies partial action by computing matching costs only for edge pixels rather than all pixels in the image. This selective computation maintains depth accuracy for structurally important regions while significantly reducing the total computational load, thereby improving processing speed.
3Measurement precision
If the system computes matching costs for all pixels, then measurement precision is improved, but productivity is reduced due to increased computation time
Solution Approach 1:
The patent segments the pixel population into edge pixels and non-edge pixels, processing only the edge pixels for full cost calculation. This segmentation maintains matching precision for critical features while reducing the total number of computations required, thereby improving productivity and enabling real-time processing.
Solution Approach 2:
The patent changes the parameter of pixel resolution by downsampling the input images before processing. This parameter change reduces the total number of pixels that need to be processed, decreasing computation time and improving productivity while maintaining matching precision through the preservation of edge structures in the downscaled images.
Data Source
AI summary
A first image and a second image of an object taken with different viewing directions are received. The first image and the second image are downscaled in a ratio of a downscale factor DF to generate a first downscaled image and a second downscaled image, respectively. An edge map is generated by detecting an edge pixel from the first downscaled image. An initial cost volume matrix is generated from the first downscaled image and the second downscaled image according to the edge map. An initial disparity estimate is generated from the initial cost volume matrix. The initial disparity estimate is refined using the initial disparity estimate to generate a final disparity set. A depth map is generated from the first image and the second image using the final disparity set.