Sub-pixel Parallax Calculation for Stereo Image Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing methods for calculating parallax from stereoscopic images often face accuracy issues due to noise and poor texture in images, especially when the parallax changing quantity is small, leading to errors in parallax calculation.
Innovation Solution
An image processing method that calculates matching costs in sub-pixel units with higher resolution, using a cost map generating process and synthesized cost calculation to improve parallax accuracy by integrating penalty costs and fitting correlation functions, thereby smoothing parallax variations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If image search processing is used to calculate parallax from stereoscopic images, then parallax can be obtained in pixel units, but the accuracy is insufficient when the parallax changing quantity is small or when noise and poor texture are present
Solution Approach 1:
The patent changes the parameter of matching cost calculation from pixel-level to sub-pixel-level precision. By calculating matching costs at sub-pixel positions (e.g., 0.5 pixel intervals) and integrating these costs, the system achieves higher parallax measurement precision while maintaining reliability under noisy conditions and poor texture scenarios.
2Measurement precision
If matching costs are calculated only at pixel units, then processing is simple and fast, but the parallax resolution is insufficient for high-precision measurements
Solution Approach 1:
The patent introduces a sub-pixel dimension to the traditional pixel-based matching cost calculation. By calculating costs at multiple sub-pixel positions (e.g., 0.0, 0.5, 1.0 pixels) and integrating these across the sub-pixel dimension, the system achieves higher resolution without requiring a complete redesign of the processing architecture.
3Measurement precision
If the parallax changing quantity between adjacent pixels is small, then the images appear similar, but this makes it difficult to distinguish correct matching and leads to calculation errors
Solution Approach 1:
The patent segments the matching cost calculation into multiple sub-pixel level calculations. Instead of calculating a single matching cost at pixel level, the system calculates costs at multiple sub-pixel positions (e.g., 0.5 pixel intervals) and integrates them. This segmentation allows better differentiation of subtle parallax changes by accumulating evidence across multiple measurement points.
Data Source
AI summary
An image processing method includes a matching cost calculating process of calculating matching costs in a unit of sub-pixels having higher resolution than first and second images by using an image of a reference area contained in the first image in which a target object is imaged and images of a plurality of comparison areas contained in the second image in which the target object is imaged, and a synthesized cost calculating process of calculating synthesized costs related to the reference area based on comparison results of values related to the plurality of matching costs calculated in the matching cost calculating process.