Infrared small targets detection and tracking based on soft morphology Top-Hat and SPRT-PMHT

This paper discusses the infrared small targets detection and tracking problem. Based on mathematical morphology, the Soft mathematical morphology filter can filters out the local catastrophe points, but at the same time the noise is also enhance, and morphology TOP-HAT filter performs better than Soft mathematical morphology filter, but it can not locate the target preciously. Combing these two methods, improved TOP-HAT Soft Morphological Filter is built to solve the target detection problem in infrared images. Towards tracking in sequential images, an Sequential Probabilistic Ratio Test method (SPRT) is introduced to the Probabilistic Multiple Hypothetic Tracking (PMHT) algorithm to form a SPRT-PMHT tracking algorithm. This new algorithm can solve the initialization and track management problem which is a puzzle in former PMHT algorithms. Besides, the post data association probability is revised to improve the precision of PMHT. Simulations with real infrared image proves that Improved TOP-HAT Soft Morphological Filter, together with round structure element, has a better detection and robustness ability and SPRT-PMH does a good job on target tracking and track management, they together compose an effective infrared small targets detection and tracking system.