Edge-Guided Multiscale Segmentation of Satellite Multispectral Imagery

作     者：Chen, Jianyu Li, Jonathan Pan, Delu Zhu, Qiankun Mao, Zhihua 

作者机构：State Ocean Adm Inst Oceanog 2 State Key Lab Satellite Ocean Environm Dynam Hangzhou 310012 Zhejiang Peoples R China Univ Waterloo Dept Geog & Environm Management Waterloo ON N2L 3G1 Canada Xiamen Univ Sch Informat Sci & Engn Key Lab Underwater Acoust Commun & Marine Informa Xiamen 361005 Peoples R China 

出 版 物：《IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING》 (IEEE Trans Geosci Remote Sens)

年 卷 期：2012年第50卷第11期

页      面：4513-4520页

核心收录：

学科分类：0808[工学-电气工程] 1002[医学-临床医学] 08[工学] 0708[理学-地球物理学] 0816[工学-测绘科学与技术] 

基　　金：National Natural Science Foundation of China [40976109, 40606040] Key Projects in the National Science and Technology Pillar Program [2008BAC42B02] R&D Special Fund for Public Welfare Industry (Oceanography) [201005011-2] China Scholarship Council 

主　　题：Edge detection multiscale segmentation object-based image analysis scale selection 

摘      要：This paper presents a new approach to multiscale segmentation of satellite multispectral imagery using edge information. The Canny edge detector is applied to perform multispectral edge detection. The detected edge features are then utilized in a multiscale segmentation loop, and the merge procedure for adjacent image objects is controlled by a separability criterion that combines edge information with segmentation scale. The significance of the edge is measured by adjacent partitioned regions to perform edge assessment. The present method is based on a half-partition structure, which is composed of three steps: single edge detection, separated pixel grouping, and significant feature calculation. The spectral distance of the half-partitions separated by the edge is calculated, compared, and integrated into the edge information. The results show that the proposed approach works well on satellite multispectral images of a coastal area.