Technical Description
An object tracking method that uses maximally stable level line segments to robustly detect and track corner points across video frames, especially on object boundaries. It combines shape-based and texture-based matching to reliably find corresponding boundary features without re-detecting them in every frame.
Problems Addressed
- Boundary Tracking Failure
- Patch Straddling Boundaries
- Depth Discontinuity Issues
- Corner Re-Detection Inconsistency
- Poor Boundary Matching
- Scale / Viewpoint Variation
Tech Features
- Level-Line Corner Detection
- Reference Boundary Segment
- Shape-Based Matching
- Texture-Based Refinement
- Boundary Feature Extraction
- Periodic Feature Update
Target Audience
- Surveillance Security Industry
- Automotive AV Sector
- Robotics Automation Industry
- Manufacturing Vision Sector
- Research & Development
Tech ID: P02-1866 TRL 4 Patent Status: Granted Available For Exclusive and Non-exclusive License
