An Adaptive Approach To Object Extraction and Tracking In Complex Image Sequences,
Abstract
Based on visual perception principles, this article studies computational models and algorithms associated with the extraction and tracking of moving targets in complex images. It analyzes and tests the effects of changing relevant model parameters on image segmentation thresholds. What is different from certain conventional algorithms is that new methods make an integrated consideration of object background conditions, visual nonlinearities, interframe correlativity and differences into a two step integrated target extraction process including three criteria and a fast optimizing procedure. Object tracking makes use of binary template matching. Experimental results are given for visible light image sequences.
Document Details
- Document Type
- Technical Report
- Publication Date
- Feb 07, 1996
- Accession Number
- ADA306424
Entities
People
- Dai Kerong
- Peng Jiaxiong
- Zhang Tianxu
Organizations
- National Air and Space Intelligence Center