THE USE OF ACTIVE CONTOURS FOR THE DETECTION OF COASTLINES IN SAR IMAGES: A MODULAR KNOWLEDGE-BASED FRAMEWORK

Over the last years, active contour methods have become a basic tool in computer vision. They have proven to be efficient for various image processing applications, like reconstruction of the edges inside images or the tracing of image features. However, when applying the basic snake technique to synthetic aperture radar (SAR) remote sensing images, the detection of edges may not be satisfactory. This is caused by the special imaging technique of SAR that may tend to produce varying-contrast edges and the commonly known speckle noise. In (Seppke et al., 2010) we proposed the use of asymmetric external energy terms to cope with these problems. In this paper we extend our approach and present a modular framework for the application of snake algorithms to SAR imagery. The main emphasis of the framework is the use of higher knowledge about the scene depicted, e.g to initialize the snake with suitable parameters. Another objective is to establish a modular designed and thus highly flexible testbed for the comparison of different active contour approaches. We present the framework’s design and preliminary results for the detection of coastlines in SAR images. The proposed framework has already proven to be a valuable tool for both, the interpretation and understanding of the results. For future projects, the framework will be used to investigate and compare the results of snakes when applied to hi-resolution SAR imagery, e.g. TerraSAR-X HR Spotlight images