Automatic pickup of suspected targets in active sonar based on spectral residual method
A method based on visual saliency spectral residual method is proposed for automatically picking up sus-pected targets out of complex and serious interferences background from the active sonar detecting output.After the conception of active sonar image built,the similarity between discovering visual saliency in natural images and picking up suspected targets in active sonar images is discussed.Based on this,the mechanism,core framework and key processing steps of the new method are explained,including preprocessing sonar images,generating sali-ency map,estimating pick-up threshold and picking up suspected targets.The results of sea trial data processing show that compared to the control method based on local relative energy analysis,the new method has a lower miss rate for known targets,and can reduce the pick-up number of bright spots by about 40%,which means a lower false alarm rate.Moreover,the processing time of this method is reduced by 37%approximately.
active sonarsuspected targetautomatic pick-upvisual saliencyspectral residual methodsonar im-agesaliency detectionpost processing
国家科技期刊平台
NSTL
万方数据
