Abstract: Small infrared moving target detection is one of the key technologies of infrared detection. And it has always been a hot issue in photoelectric detection, especially for small infrared moving target under clutter background. The problem has not yet been solved. In view of the lack of high quality data in current research, we adopt infrared imaging equipment installed on a small UAV platform to capture image sequences as backgrounds. The synthetic targets are embedded in backgrounds properly. We build a semi-synthetic dataset for small infrared moving target detection under clutter background. Various conditions are set in image capturing, including relative height (up looking, head up looking and down looking), scenes (vegetation, water and building), platform motion, weather, time, etc. The imaging platform motion and the synthetic target motion are considered simultaneously to guarantee that the semi-synthetic image sequences as close as possible to the real application scenario. In addition, we vary the target characteristics in synthetizing, including shape, intensity, and motion. Our dataset contains 350 image sequences, 150185 images. The annotation file provided the targets positions in images. The dataset can be used in small infrared moving target detection and tracking researches.
Keywords: Small infrared moving target; Airborne detection platform; Clutter background; Target detection; Image sequence