A bio-inspired spatiotemporal contrast operator for small and low-heat-signature target detection in infrared imagery

Thermal infrared imaging is a promising modality for long-range small target detection. However, low target contrast, high background clutter and sensor noise are some of the key challenges that need to be resolved efficiently for robust detection performance. The spatiotemporal processing in the early stages of the visual pathway of small flying insects has a remarkable ability to simultaneously address such challenges. The first stage of the early visual system corresponds to the adaptive temporal filtering mechanisms of photoreceptor cells. This stage improves the signal-to-noise ratio, enhances target background discrimination and compresses the signal bandwidth. The second stage pertains to the adaptive spatiotemporal filtering in the lamina monopolar cells. This stage removes spatiotemporal redundancy and enhances target contrast. In this paper, we explore such two-stage bio-processing to simultaneously suppress clutter and enhance the contrast of small low-heat-signature targets in real-world infrared imagery. We also propose a simple and efficient spatial contrast operator called center–surround total differential index for target region segmentation. Small moving target detection experiments on real-world high-bit-depth infrared video sequences show that the proposed method significantly outperforms the state-of-the-art spatial and spatiotemporal infrared small target detection methods. Specifically, our method resulted in 59% better detection rate (at 10^{- 5} false alarm rate) than the best competing method. Our results show that the bio-inspired spatiotemporal preprocessing is an excellent tool for significantly improving the performance of existing long-range infrared target detection techniques.