3D Model Generation via Line Pattern Imaging Association
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques face challenges in accurately associating line segments in images, especially when they are similar or densely populated, which hinders the generation of precise three-dimensional geometric models for mixed reality and robot navigation applications.
Innovation Solution
An image processing apparatus and method that detects line segments, sets reference lines, derives pattern arrays from pixel value changes along these lines, and compares these patterns across images to associate line segments effectively, even in cases of similar or densely populated segments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual association of line segments is used, then association accuracy is improved for small numbers of line segments, but processing time and labor increase significantly as the number of line segments increases
Solution Approach 1:
The system automatically extracts feature points and performs line segment association without requiring manual intervention. The automatic association process uses algorithmic methods to match line segments across multiple images, eliminating the need for manual association while maintaining high accuracy even for large numbers of line segments.
Solution Approach 2:
The invention changes the approach from manual visual association to automated parameter-based association by extracting feature points and computing geometric relationships. This transforms the association task into a computational problem that can be solved automatically using mathematical parameters and algorithms.
2Productivity
If line segment-specific information independent of observation position is assigned, then automatic association efficiency is improved, but the complexity of extracting and processing this information increases
Solution Approach 1:
The system extracts specific feature points from line segments that have invariant characteristics across different observation positions. By isolating these distinctive feature points, the system creates a simplified representation that can be automatically associated without requiring complex processing of the entire line segment data.
Solution Approach 2:
Instead of processing the entire line segment uniformly, the invention focuses on local distinctive feature points along the line segment. These local features carry the essential identification information needed for association, while the rest of the line segment data can be processed more simply.
3Ease of manufacture
If existing line segment association techniques are used, then processing is simplified, but association accuracy deteriorates when line segments are similar or densely populated
Solution Approach 1:
The system segments the line segment association task into multiple stages: first extracting distinctive feature points from line segments, then using these features for association. This segmentation allows the system to handle similar and densely populated line segments by focusing on their distinctive features rather than treating them as uniform structures.
Solution Approach 2:
The invention introduces asymmetric feature extraction that identifies unique characteristics in line segments regardless of their orientation or position. By extracting asymmetric distinctive features, the system can differentiate between similar line segments and establish accurate associations even in dense configurations.
Data Source
AI summary
An image processing apparatus comprising: an input unit configured to input a plurality of images obtained by capturing a target object from different viewpoints; a detection unit configured to detect a plurality of line segments from each of the plurality of input images; a setting unit configured to set, for each of the plurality of detected line segments, a reference line which intersects with the line segment; an array derivation unit configured to obtain a pattern array in which a plurality of pixel value change patterns on the set reference line are aligned; and a decision unit configured to decide association of the detected line segments between the plurality of images by comparing the pixel value change patterns, contained in the obtained pattern array, between the plurality of images.