Infrared Small Target Detection Based on Resampling-Guided Image Model

Abstract

Infrared target detection is of great significance to various critical fields, however, it is still a crucial problem to robustly and quickly detect infrared small targets under complicated circumstances. In this letter, a novel detection method based on resampling-guided-image model is proposed. Initially, a new infrared image model based on guided image filtering is introduced as an improved supplant for normally used infrared patch-image (IPI) model. Our model approximates the effect of IPI model in strengthening the low-rankness of background components and provides a favorable presuppression to the noise components. Furthermore, the partial sum of singular values and nonnegativity are adopted to constrain the background image and the target image, respectively. Besides, the alternating direction method of multipliers is adopted to efficiently solve the model. A series of experiments show that the proposed algorithm has a more robust and efficient detection performance compared with other baselines.