Stereo Matching Search Range Prediction via Edge Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current stereo matching methods are inefficient in terms of time and resource usage, particularly in devices with limited battery capacity and computing resources, and struggle to maintain accuracy while reducing the search range for disparity calculations.
Innovation Solution
The method involves edge detection to determine edge lines and feature points, sampling reference pixels, predicting a reduced search range based on disparities of reference pixels, and applying correction factors to enhance accuracy and efficiency in stereo matching.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a full search range is used for stereo matching of all pixels, then matching accuracy is maintained, but computational time and resource usage increase significantly
Solution Approach 1:
The patent divides the image into multiple regions based on edge detection results, processing different regions with different search range strategies. Edge regions use larger search ranges while non-edge regions use smaller search ranges, thereby reducing overall computational time while maintaining matching accuracy in critical areas
Solution Approach 2:
The patent applies different search range sizes to different spatial locations based on local image characteristics. By identifying edge lines and feature points, the system dynamically adjusts search range quality - using full search ranges only where needed (edges) and reduced ranges elsewhere, resolving the contradiction between global accuracy and local efficiency
2Productivity
If the search range is reduced to improve processing speed, then computational efficiency increases, but matching accuracy deteriorates
Solution Approach 1:
The patent performs edge detection and identifies feature points before conducting stereo matching. This preliminary action allows the system to pre-determine which regions require full search ranges and which can use reduced ranges, ensuring accuracy is maintained in critical areas while improving overall processing speed
Solution Approach 2:
The patent dynamically adjusts search range size based on local image features rather than using a fixed search range. The search range becomes a dynamic parameter that adapts to edge lines and feature points, allowing the system to optimize the balance between processing speed and matching accuracy for each specific region
3Measurement precision
If edge detection and feature point sampling are performed, then search range prediction accuracy improves, but additional computational steps are required
Solution Approach 1:
The patent extracts only the most critical elements (edge lines and salient feature points) from the image rather than processing all pixels. By taking out only the essential structural information needed for search range prediction, the system achieves accurate predictions without the complexity of processing the entire image
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Provided is a stereo matching method and apparatus. A stereo matching method includes acquiring detection information based on an edge detection of a first image, determining an edge line and a feature point based on the detection information, sampling a reference pixel from the edge line and the feature point, determining a polygon with the reference pixel as a vertex, predicting a search range comprising a disparity of a pixel within the polygon based on a disparity of the reference pixel, and performing stereo matching of the pixel based on the search range.