Stereo Imaging Disparity Map Edge Refinement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing imaging systems that generate depth maps through stereo-matching face a tradeoff between depth map quality and computational complexity, with global methods being intractable and local methods sacrificing quality for lower complexity.
Innovation Solution
The proposed imaging system employs edge refinement to improve disparity map quality without increasing computational complexity, using edge-detecting, disparity-estimating, and edge-refining modules to refine edge coordinates and propagate disparity levels, thereby enhancing map quality while reducing complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If global methods use larger neighborhood for smoothness term to obtain better boundaries, then depth map quality is improved, but computational complexity becomes intractable
Solution Approach 1:
The patent segments the image processing into distinct stages: initial disparity estimation using local methods, edge detection to identify boundary locations, and edge refinement to improve quality only at critical regions. This segmentation allows the system to avoid global optimization while still achieving good boundary quality where needed.
Solution Approach 2:
The patent applies local quality by performing edge refinement only at detected edge locations rather than globally across the entire image. The edge refinement module uses local gradient information and matching around detected edges to improve disparity accuracy at boundary regions while leaving non-edge regions unchanged, thus improving quality where it matters most without incurring global computational costs.
2Device complexity
If local methods are used to reduce computational complexity, then computational complexity is lowered, but depth map quality deteriorates
Solution Approach 1:
The patent performs preliminary edge detection on the input image pair before disparity estimation. By identifying edge locations in advance, the system can then apply edge refinement after initial disparity estimation to correct boundary inaccuracies. This preliminary action allows the system to use computationally efficient local methods for the main disparity estimation while still achieving good boundary quality through the subsequent refinement step.
Solution Approach 2:
The edge detection and edge refinement modules act as intermediaries between the initial local disparity estimation and the final depth map output. These intermediary steps detect and correct boundary inaccuracies that would otherwise be present in simple local methods, effectively bridging the quality gap without requiring full global optimization.
Data Source
AI summary
A imaging system and method is disclosed. In one aspect, the system includes a first edge-detecting module configured to detect edge coordinates in the first image, a first disparity-estimating module configured to obtain a first estimated disparity map of the first image relative to the second image, and a first edge-refining module configured to refine edge coordinates in the first estimated disparity map using the edge coordinates in the first image to obtain a first refined disparity map. The imaging system and method improve the quality of a disparity map and control the complexity of stereo-matching.