Infrared target detection and false alarm elimination based on multi-feature fusion decision

摘要

In this paper a multi-feature fusion decision algorithm based on the idea of lingering decision is proposed. Firstly, the algorithm segments the infrared images by iterative OTSU segmentation method and detects the edge by morphological processing. Then, it extracts multiple features of the ted targets on the basis of the combination of segmentation and edge result. Finally, an integrative classifier is constructed to distinguish the real targets from the false alarm by setting up appropriate fusion criteria. The test data detection demonstrates the validity of the algorithm.