A Fast Algorithm for Coastline Detection from Remote Sensing Images Based Variational Level Set Methods

摘要

In this paper, we present a fast algorithm for coastline detection from remote sensing images based variational level set methods. Our algorithm combines two variational level set methods together, which are a fast algorithm for level set based optimization and the level set evolution without re-initialization. When applying the former algorithm to the Chan-Vese model, it can get a contour that is the outline of the coastline rapidly. Then using the latter model, we evolve the contour as the initial curve. It can sweep out the isolated curve on the land part and preserve the coastline has been detected. The resulting evolution of the contour is the coastline that we detect finally. Our algorithm resolves a conflict between the speed and the accuracy for coastline detection. It can achieve good results in the segmentation in a short time and has the ability to detect the objects from noise images. Several experiments have been conducted on the real images and the numerical results exhibit the proposed algorithm is effective and has good segmentation performance for the coastline detection from remote sensing images.