A Region-Based Image Fusion Algorithm for Detecting Trees in Forests

Abstract

Multi-sensor image fusion has been widely used in many areas. In this paper, a region-based method using NSCT and fuzzy logic is proposed to enhance trees' contour and enrich the information in images. Firstly, the required source images are segmented, which are then decomposed using the NSCT transform. In low frequency domain, the fuzzy fusion method is used. And in high frequency domain, a region-based rule is proposed to highlight the objects' detail within images. Finally, the fused images are obtained by an inverse NSCT transform. Nine pieces of objective criteria and computation time are adopted to evaluate the fusion result. Moreover 30 groups of collected images are utilized to prove the robustness of the proposed method. The results show that the algorithm could provide fused images which have a higher visual effect, combine more information and consume less time. Furthermore, this process could better capture the contour and texture of the trees.