Dim small target detection in single frame complex background

Abstract

High precision detection of small targets in complex background is a challenging task, which has not been well resolved. In this paper, the improve sparse representation(ISR) algorithm is proposed based on the characteristics of passive millimeter wave imaging as well as the difference of the priori characteristics between the small target and background clutter. Firstly, the algorithm constructs a over-complete dictionary based on the content of the image itself, and then improves the original sparse representation method to complete precise classification of target and background dictionaries. After background suppression and target enhancement, we can easily extract the target. The millimeter wave images of different scenes are detected and the results show that compared with some other mainstream algorithms, the ISR algorithm has lower false alarm rate, higher detection accuracy and stronger robustness.