SAR image target detection based on multi-scale auto-convolution variance saliency

Aimed at the target detection problems of strong echoes in SAR images under complex background, an adaptive target detection algorithm is proposed on the basis of the multiscale auto-convolution variance saliency (MSAVS). First, with the calculation of MSAVS, the variance saliency map is obtained by using the presented algorithm. Second, an auto-selecting-threshold detector is constructed according to the complexity of SAR image. Finally, the methods of choosing the window size of MSAVS filter and the scale parameter of multi-scale auto-convolution are analysed, and the salient objects detection in SAR image was completed. Experiment results show that by using the algorithm presented in complex scene, the salient object consistent comparatively with human visual sense could be effectively detected.