A Local Contrast Method for Infrared Small-Target Detection Utilizing a Tri-Layer Window

Jinhui Han, Saed Moradi, Iman Faramarzi, Chengyin Liu, Honghui Zhang, Qian Zhao

IEEE Geoscience and Remote Sensing Letters · 2019 · 455 citations · 24 references

Concepts

TL;DR

Local contrast is effective for infrared small‑target detection, but existing methods lack targeted enhancement before contrast calculation and are susceptible to noise. The study proposes a multiscale tri‑layer local contrast measure (TLLCM) framework for infrared small‑target detection. TLLCM uses a tri‑layer window (core, reserve, surrounding) where a Gaussian‑matched filter enhances the core, then computes multiscale contrast between the enhanced core and surrounding background, and finally extracts targets via an adaptive threshold. Experiments demonstrate that TLLCM outperforms several existing infrared small‑target detection algorithms.

Abstract

Local contrast has been proved efficient for infrared (IR) small-target detection. However, current algorithms do not enhance true target purposefully before local contrast calculation and may easily be disturbed by noises. In this letter, a new detection framework named multiscale tri-layer local contrast measure (TLLCM) is proposed. First, a tri-layer filtering window is proposed, and it consists of a core layer, a reserve layer, and a surrounding layer. The idea of a matched filter is adopted, and a Gaussian filtering will be performed on the core layer to enhance true target purposefully according to the target shape. Then, the multiscale TLLCM of the central pixel of the window will be calculated between the enhanced core and the surrounding local background. Finally, the target can be extracted by an adaptive threshold. Experimental results show that the proposed method can achieve better detection performance than some existing algorithms.

References

24