Stereo Matching Disparity Derivation for Weak Texture Objects
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional stereo matching methods fail to accurately derive disparity values for objects with weak texture, as costs are affected by surrounding features with strong texture, leading to inaccurate distance measurements.
Innovation Solution
A disparity value deriving device that calculates costs for candidates on an epipolar line, changes costs exceeding a threshold to a higher value, synthesizes costs for corresponding regions, and derives disparity values based on the smallest synthesized cost, effectively isolating the impact of weak texture objects from strong texture surroundings.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If costs of all pixels including surrounding strong texture regions are aggregated to derive disparity for weak texture objects, then more disparity information is available, but the costs for weak texture objects are contaminated by costs from strong texture surroundings
Solution Approach 1:
The patent segments the cost information by introducing a confidence map that divides pixels into reliable (strong texture) and unreliable (weak texture) regions. This segmentation allows the system to selectively aggregate costs only from reliable regions when calculating disparity for weak texture objects, preventing contamination from strong texture surroundings while maintaining sufficient cost information for accurate disparity derivation.
Solution Approach 2:
The patent applies local quality by making the cost aggregation process adaptive to local image characteristics. The confidence map evaluates texture strength locally for each pixel, and the cost synthesis selectively weights or excludes costs from neighboring pixels based on their local texture quality. This ensures that weak texture regions receive cost information only from other weak texture regions with similar characteristics, improving measurement precision.
2Device complexity
If conventional stereo matching is used for weak texture regions, then processing is simple, but sufficient effect cannot be obtained and accurate disparity cannot be derived
Solution Approach 1:
The patent performs preliminary action by generating a confidence map before the disparity calculation process. This confidence map pre-identifies which pixels have strong texture (reliable) and which have weak texture (unreliable). By preparing this classification in advance, the system can then efficiently aggregate costs only from appropriate regions during the disparity calculation, achieving accurate results for weak texture objects without excessive processing complexity.
Solution Approach 2:
The patent introduces a confidence map as an intermediary element that mediates between the raw image data and the disparity calculation process. This intermediary structure stores texture quality information and guides the cost aggregation process, enabling the system to handle weak texture regions accurately while maintaining overall processing efficiency. The confidence map acts as a bridge that translates local texture characteristics into global disparity calculation decisions.
Data Source
AI summary
A disparity value deriving device includes a calculator configured to calculate costs of candidates for a corresponding region in a comparison image that corresponds to a reference region in a reference image, based on luminance values of the regions. The device also includes a changer configured to change a cost exceeding a threshold to a value higher than the threshold; a synthesizer configured to synthesize a cost of a candidate for a corresponding region for one reference region after the change and a cost of a candidate for a corresponding region for another reference region after the change; and a deriving unit configured to derive a disparity value based on a position of the one reference region and a position of the corresponding region in the comparison image for which the cost after the synthesis is smallest.