Region of interest extraction in remote sensing images by saliency analysis with the normal directional lifting wavelet transform

Abstract

Region of interest (ROI) extraction techniques based on saliency comprise an important branch of remote sensing image analysis. In this study, we propose a novel ROI extraction method for high spatial resolution remote sensing images. High spatial resolution remote sensing images contain complex spatial information, clear details, and well-defined geographical objects, where the structure, edge, and texture information has important roles. To fully exploit these features, we construct a novel normal directional lifting wavelet transform to preserve local detail features in the wavelet domain, which is beneficial for the generation of edge and texture saliency maps. We also improve the extraction results by calculating the amount of self-information contained in the spectra to obtain a spectral saliency map. The final saliency map is a weighted fusion of the two maps. Our experimental results demonstrate that the proposed extraction algorithm can eliminate background information effectively as well as highlighting the ROIs with well-defined boundaries and shapes, thereby facilitating more accurate ROI extraction.