Multi-image Color Refinement for Disparity Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing multi-camera vision-based depth estimation systems face challenges in accurately matching image features due to noise and color biases between cameras, particularly in areas with gradual changes in color or intensity, leading to incorrect matches and loss of intrinsic information.
Innovation Solution
A method for multi-image color-refinement and disparity map generation, where images are iteratively aligned and refined to adjust pixel colors based on corresponding pixels in other images, using techniques like weighted non-linear color space warping, to enhance spatial alignment and disparity estimation without normalizing for color or lighting changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If image patches or neighborhoods are used to compare regions around points, then noise is mitigated, but color biases between cameras remain uncorrected leading to incorrect matches
Solution Approach 1:
The patent applies color refinement by adjusting pixel values in one image to match corresponding pixels in another image. This involves changing the color parameters (R, G, B values) of pixels based on disparity information and color differences between cameras, thereby correcting color biases while preserving matching information in gradual gradient areas.
2Reliability
If methods like Normalized Cross-Correlation or Census transform are used to handle color or lighting changes, then matching is improved, but intrinsic information in areas of limited texture is filtered and discarded
Solution Approach 1:
The patent applies local color refinement adjustments based on disparity information. Instead of globally normalizing colors, it performs localized pixel value adjustments in gradual gradient areas based on corresponding pixel colors in the other image. This preserves local gradient information while correcting color biases, allowing matching in areas with limited texture.
3Productivity
If pixel-wise or patch-based matching is used, then matching speed is maintained, but gradually changing image areas cannot be matched
Solution Approach 1:
The patent performs color refinement as a preliminary step before final matching. By pre-adjusting pixel colors in one image to match the other image's color characteristics, it prepares the images for more effective matching. This preliminary color correction enables both pixel-wise speed and the ability to match gradual gradient areas.
Data Source
AI summary
Systems, methods, and computer-readable media to improve multi-image color-refinement operations are disclosed for refining color differences between images in a multi-image camera system with application to disparity estimation. Recognizing that corresponding pixels between two (or more) images of a scene should have not only the same spatial location, but the same color, can be used to improve the spatial alignment of two (or more) such images and the generation of improved disparity maps. After making an initial disparity estimation and using it to align the images, colors in one image may be refined toward that of another image. (The image being color corrected may be either the reference image or the image(s) being registered with the reference image.) Repeating this process in an iterative manner allows improved spatial alignment between the images and the generation of superior disparity maps between the two (or more) images.