Target detection in maritime search and rescue using SVD and frequency domain characteristics

Abstract

A target detection method based on SVD and frequency domain characteristics in maritime search and rescue is proposed. The method can select maritime search and rescue targets from textured sea scenes. Firstly, intensity component is obtained and is decomposed using SVD. The first 80% singular components are combined to represent the dominant pattern of input image. Secondly, spectral residual approach and phase spectrum approach are applied to the difference between intensity component and corresponding dominant pattern. The resulting saliency maps show the salient region in maritime search and rescue visual scene and there by target detection performed.