Infrared small target detection based on saliency and difference of Gaussian

Abstract

The challenge of detecting the infrared small target in complex scenes is how to enhance the target area and suppress the interference of the background and noise. We present an infrared small target detection method based on saliency and difference of Gaussian. By calculating the spectral residuals of the original infrared images to obtain the saliency map, and a gray difference map is computed by adopting the difference of Gaussian. By normalizing and fusing these maps, the final feature response map is obtained, which can restrain the background noise and enhance the targets. Finally, the adaptive threshold segmentation method is used to detect small targets from the feature response map. Experimental result indicates that the proposed method achieves better image processing effect than traditional detection algorithm, works well under the interference of background and noise as well as detects the target accurately under different complex backgrounds.