Lake shoreline detection and tracing in SAR images using wavelet transform and ACM method

Qulin Tan,Zhengjun Liu,Zhou Fu,Jiping Hu

DOI: https://doi.org/10.1109/IGARSS.2005.1526655

2005-01-01

Abstract:This paper presents an approach to lake shoreline detection and tracing in SAR images using wavelet transform and ACM (active contour model) method. In this method, we take SAR speckle noise as weighted impact function and take image edges as an aggregation of neighboring points with similar singularity. Based on the above imaginary model, we design the following scheme. Firstly, a fast dyadic wavelet transform is performed in SAR images. Then, we separate edge points from noise points using the different propagation properties of singularities of noise and edge across the multi-scale dyadic wavelet transform. Under the space congruity condition, the common edge points are ascertained the true edge points. Finally, we extract the edges by connecting with these detected edge points using ACM method based on gradient vector flow. The experimental data are Radarsat scanSAR images of DongTing Lake area, China, acquired on Aug. 24, 1998, which resolution is 50 meters. The result shows that the developed method can accurately and efficiently detect and trace out the lake shorelines in the Radarsat scanSAR images and at the same time effectively suppress the noise interference, even in the SAR images having serious noise and complex edges, such as zigzag lake shorelines. © 2005 IEEE.