Region-of-interest extraction based on spectrum saliency analysis and coherence-enhancing diffusion model in remote sensing images

Abstract

Saliency analysis, which is a fast and efficient method for extracting Regions of Interest (ROIs), has been applied in the field of remote sensing image analysis. In this paper, a novel saliency analysis model is proposed using spectrum saliency analysis (SSA) and coherence-enhancing diffusion model (CED). In SSA, the probability of each pixel intensity value is counted to get the one-dimensional histogram for each band, then information content of each band is calculated according to the one-dimensional histogram and generate spectrum saliency map after the information content weighted fusion of each band. In CED, coherence-enhancing diffusion model is introduced for spectrum saliency map that aims to smooth internal ROIs and eliminate background interference to improve the precision, accuracy and completeness of the resulting saliency map. Finally, ROIs are segmented from the saliency maps of original images by an adaptive threshold segmentation algorithm. Several experiments were conducted to evaluate the overall performance of the proposed model compared with the other eight outstanding models qualitatively and quantitatively. The experimental evaluations show that the proposed model outperforms the relevant outstanding models.