3D Contour Recognition Using Stereo Image Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional 3D contour recognition and mesh generation systems face challenges with slow and inaccurate stereo image matching, particularly due to self-occlusion issues, which result in erroneous depth calculations and reduced processing speed, especially on lower-powered devices.
Innovation Solution
A system comprising two cameras and a processor that extracts 2D contours from images, matches points to obtain matched and mismatched contour points, and reconstructs the object's shape, using a method that processes contours instead of images, enabling faster and more accurate stereo image matching, reaching 60 fps or higher on a Mobile ARM CPU.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional stereo image matching is used, then depth information can be obtained, but processing speed is slow and accuracy is reduced due to self-occlusion issues
Solution Approach 1:
The patent segments the image processing task by extracting only contour information from images rather than processing entire images. This segmentation of visual data into essential boundary elements enables faster processing while maintaining depth measurement capability through contour point matching between stereo images
Solution Approach 2:
The patent extracts contour points from stereo images as the essential feature for depth calculation, taking out only the necessary boundary information rather than processing all pixel data. This extraction approach eliminates self-occlusion issues by focusing on visible contours while maintaining accurate depth measurement through geometric reconstruction
2Reliability
If conventional image-based 3D reconstruction is used, then complete image data is processed, but computational power requirements are high and processing speed is slow
Solution Approach 1:
The patent extracts only contour points from images, removing unnecessary pixel data while retaining essential shape information for 3D reconstruction. This extraction reduces computational power requirements significantly while maintaining reconstruction reliability through accurate contour point matching and geometric modeling
Solution Approach 2:
The patent changes the data representation parameter from full image pixel arrays to contour point coordinates. This parameter transformation reduces data dimensionality and computational complexity while preserving the geometric information needed for accurate 3D reconstruction
Data Source
AI summary
A system for computer vision is disclosed. The system may comprise a processor and a non-transitory computer-readable storage medium coupled to the processor. The non-transitory computer-readable storage medium may store instructions that, when executed by the processor, cause the system to perform a method. The method may comprise obtaining a first and a second images of at least a portion of an object, extracting a first and a second 2D contours of the portion of the object respectively from the first and second images, matching one or more first points on the first 2D contour with one or more second points on the second 2D contour to obtain a plurality of matched contour points and a plurality of mismatched contour points, and reconstructing a shape of the portion of the object based at least in part on at least a portion of the matched points and at least a portion of the mismatched contour points.