Comparison of different edge detection algoriths for shoreline boundaries extraction using radatsat SAR imagery

Abstract : this paper presents work on extracting shoreline boundaries from radarsat synthetic aperture radar [sar] imagery for shoreline mapping. shoreline boundaries extraction from sar data belongs to the boundaries detection problem in the field of image processing, in which image segmentation and edge detection algorithm are two conventional techniques to the boundaries detection. sar data is generally interrupted by speckle. the presence of speckle makes edge detection algorithm in sar data difficult and image segmentation inadequate. for the purpose of this study, the edge detection algorithm was utilized. several edge detection algorithms (sobel, prewitt and canny) were applied on radarsat sar imagery and compared for their ability on extraction of shoreline boundaries along the kuala terengganu shoreline, malaysia. the gamma filter was used for removing speckle on radarsat sar image and gamma filter size 7x7 provided the best result. in this study, we proposed a combination of canny edge detection algorithm with thinning algorithm for the best extracting concave shoreline from radarsat sar imagery. it can be concluded that canny algorithm has capability for shoreline extraction from radarsat sar imagery without involving further processing analysis and it can be determined the true water/land boundaries.