Narrow Band Multi-Region Level Set Method for Remote Sensing Image

Fang Jiang-xiong,Tu En-mei,Yang Jie,Jia Zhen-hong,Nikola Kasabov
DOI: https://doi.org/10.3964/j.issn.1000-0593(2011)11-3001-05
2011-01-01
Spectroscopy and spectral analysis
Abstract:Massive redundant contours happen when the classical Chan-Vese (C-V) model is used to segment remote sensing images, which have interlaced edges. What's more, this model can' t segment homogeneous objects with multiple regions. In order to overcome this limitation of C-V model, narrow band multiple level set method is proposed. The use of N-1 curves is required for the segmentation of N regions and each curve represents one region. First, the level set model to establish an independent multi-region region can eliminate the redundant contours and avoids the problems of vacuum and overlap. Then, narrow band approach to level set method can reduce the computational cost. Experimental results of remote image verify that our model is efficient and accurate.
What problem does this paper attempt to address?