River extraction of SAR images via active contours driven by adaptive global fitting energies

Affiliation of Author(s):电子信息工程学院

Journal:Int. J. Remote Sens.

Abstract:This article proposed a novel active contour model based on adaptive global fitting energies to achieve accurate river extraction of synthetic aperture radar (SAR) images. First, inspired by the mean local fitted image in the local image fitting energy model, the mean global fitted image is defined. Second, to adapt to intensity characteristics of SAR images, we define another median global fitted image which can suppress the effect of interference regions. Then the energy functional of the proposed model is obtained by minimizing the differences between the original image and above two kinds of global fitted images. Moreover, the minimum absolute differences of pixel intensities inside object and background regions are utilized to adaptively adjust energies of object and background regions. The experimental results of river extraction of SAR images demonstrate that the proposed model can extract rivers of SAR images accurately and efficiently and has advantages in extraction performance and extraction efficiency. © 2018, © 2018 Informa UK Limited, trading as Taylor & Francis Group.

ISSN No.:0143-1161

Translation or Not:no

Date of Publication:2018-10-18

Co-author:韩斌

Correspondence Author:wyq

Date of Publication:2018-10-18