Disparity Map Building Using Guide Nodes for Stereo Vision
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing stereo vision technologies face challenges in generating disparity maps quickly and accurately, particularly for real-time applications like autonomous driving, due to high computational costs and noise issues with existing methods such as Block Matching and Semi-Global Block Matching, which also limit sub-pixel level accuracy and introduce waveform distortions.
Innovation Solution
The use of guide nodes to build disparity maps, which involves selecting a guide node pixel and surrounding pixels, performing a full range search for the guide node pixel peak location, and calculating shift offsets for surrounding pixels using block matching operations in a local range, reducing computational complexity and noise, and enabling sub-pixel level accuracy without post-processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If Block Matching is used to build disparity maps, then processing speed is improved, but matching accuracy deteriorates resulting in high number of mismatches
Solution Approach 1:
The image is divided into guide node pixels and surrounding pixels. Guide node pixels serve as reference points for block matching operations, while surrounding pixels use the guide node's disparity information. This segmentation allows different processing strategies for different pixel types, improving both speed and accuracy.
Solution Approach 2:
Guide node pixels act as intermediaries between full search and block matching operations. The guide node undergoes full search for accurate disparity determination, then serves as a mediator to provide disparity information to surrounding pixels through block matching, eliminating the need for expensive full search on every pixel.
2Measurement precision
If Semi-Global Block Matching is used to build disparity maps, then matching accuracy is improved, but processing speed deteriorates making it complex and slow
Solution Approach 1:
The processing area is segmented into guide nodes and surrounding pixels. Only guide nodes undergo computationally intensive full search operations, while surrounding pixels use faster block matching with the guide node's disparity as reference. This reduces overall computational complexity while maintaining accuracy.
Solution Approach 2:
Instead of performing full search on all pixels (excessive action), the method performs full search only on guide nodes (partial action). The surrounding pixels use the guide node's disparity information through block matching, which is sufficient for most cases and dramatically reduces processing time while maintaining acceptable accuracy.
3Productivity
If pixel level accuracy is used to build disparity maps, then processing speed is improved, but measurement precision deteriorates resulting in waveform distortions
Solution Approach 1:
The method changes the search range parameter dynamically. Guide nodes use full range search for maximum accuracy, while surrounding pixels use local range search centered at the guide node's disparity location. This parameter change enables sub-pixel level accuracy without requiring post-processing, eliminating waveform distortions while maintaining processing efficiency.
Data Source
AI summary
An apparatus including an interface and a processor. The interface may be configured to receive pixel data. The processor may be configured to generate a reference image and a target image from the pixel data, perform disparity operations on the reference image and the target image and build a disparity map in response to the disparity operations. The disparity operations may comprise selecting a guide node from the pixel data comprising a pixel and a plurality of surrounding pixels, determining a peak location for the pixel by performing a full range search, calculating a shift offset peak location for each of the surrounding pixels by performing block matching operations in a local range near the peak location and generating values in a disparity map for the pixel data in response to the peak location for the pixel and the shift offset peak location for each of the surrounding pixels.