Infrared small target detection based on local image alignment in complex background

Abstract

Infrared (IR) small target detection is widely used in both civilian and military security fields. However, IR small target detection in complex backgrounds faces many challenges. For example, small targets often occupy only a few pixels in IR images, lacking texture and contour features. IR images are seriously disturbed by clutter and noise, which results in the target being easily submerged. Therefore, it is difficult to achieve a low false alarm rate and a high detection rate at the same time. In this paper, a small target detection method based on local image alignment is proposed. First, the thermal IR imaging system is combined with the range-gated technology. The range-gated technology can be used to shield the background of the non-gated area. Second, the continuous frames in the sequence are accumulated to suppress noise. The movement of the target will form a trailing line in the accumulated image, and the trailing line contains the motion information of the target. Third, the trailing line is detected and extracted. Each trailing line corresponds to a suspected target area, including the real target and edges in the background, and the motion information of the suspected target can be calculated from the length and direction of the trailing line. Then, according to the motion information, the local images of the suspected target in a certain number of consecutive frames are aligned and accumulated. In the accumulated image, the real small target will shrink to a bright spot, and the signal-to-noise ratio will be significantly improved, while the background edge still appears as a line. Finally, the target is extracted from the accumulated image.