A Novel Biologically Inspired Target Detection Method based on Saliency Analysis in SAR Imagery

Abstract Saliency Object Detection (SOD) models driven by the biologically-inspired Focus of Attention (FOA) mechanism can obtain highly accurate saliency maps. However, their application in the high-resolution Synthetic Aperture Radar (SAR) images faces some intractable problems due to complex background. In this paper, we propose a novel hierarchical self-diffusion saliency (HSDS) method for detecting vehicle targets in large scale SAR images. To reduce the influence of cluttered returns in saliency analysis, we learn a weight vector from the training set to capture the optimal initial saliency of the superpixels during saliency diffusion. Considering the background objects have multiple sizes, the saliency analysis is implemented in multi-scale space, and a saliency fusion strategy then is employed to integrate the multi-scale saliency maps. Simulation experiments demonstrate that our proposed method with these improvements can achieve more accurate detection and lead to less false alarms, compared to benchmark approaches.