3D Reconstruction Using Line-Stripe Coding for Dynamic Scanning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing structured light three-dimensional reconstruction technologies face challenges with low scanning efficiency due to limited code element coding capacity and accuracy issues in spatial coding, particularly in dynamic scenes.
Innovation Solution
A method utilizing line segment stripe coding with randomly distributed code elements, enhancing coding capacity and accuracy by determining candidate and target code elements through epipolar line feature points and triangulation, improving scanning efficiency and data density.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If temporal coding is used to project multiple frames of patterns, then three-dimensional reconstruction can be achieved, but high frame rate scanning cannot be realized and applicable scenes are limited
Solution Approach 1:
The patent uses periodic projection of structured light patterns (multiple frames) to encode depth information temporally, allowing accurate 3D reconstruction while maintaining the ability to handle dynamic scenes through high-speed sequential projection
Solution Approach 2:
The patent changes the temporal parameter by projecting patterns at high frame rates and uses phase-shifting techniques to encode depth information in the temporal domain, resolving the contradiction between reconstruction accuracy and scanning speed
2Ease of operation
If circular code element coding method is used with relative displacement relationship, then decoding can be performed, but code element coding capacity is small and only sparse code element points can be reconstructed
Solution Approach 1:
The patent transitions from one-dimensional circular code elements to two-dimensional line segment stripe patterns, dramatically increasing the coding capacity by utilizing additional spatial dimensions and allowing dense sampling of code elements across the entire field of view
Solution Approach 2:
The patent changes the geometric parameters of code elements from small circular shapes to extended line segments with varying orientations and positions, exponentially increasing the number of distinguishable code elements and enabling dense 3D point cloud reconstruction
3Productivity
If image blocks are used for matching, then three-dimensional data can be acquired, but matching accuracy is relatively low
Solution Approach 1:
The patent segments the image into individual code element units rather than using large image blocks, allowing precise matching of each code element's position and orientation while maintaining high computational efficiency through localized processing
Solution Approach 2:
The patent uses multiple geometric parameters (position, orientation, length, width) of line segment stripe code elements for matching instead of simple image block correlation, significantly improving matching accuracy through multi-parameter constraint optimization
Data Source
AI summary
Disclosed in the present disclosure are a method, apparatus and system for three-dimensional reconstruction. The method includes: projecting a first image onto a surface of a measured object to obtain a second image; determining, in the first image, a plurality of candidate code elements corresponding to any code element in the second image, and determining, from the plurality of candidate code elements, a target code element corresponding to the any code element to complete matching between the code elements in the first image and the second image; and determining three-dimensional coordinates of the code elements in the second image based on a matching relationship between the code elements in the first image and the second image and predetermined three-dimensional coordinates of the code elements in the first image, and determining three-dimensional coordinates of the measured object based on the three-dimensional coordinates of the code elements to complete three-dimensional reconstruction.