Skip to Main Content
This paper presents a new image segmentation algorithm based on edge-preserving watershed for the auto-segmentation of SAR sea ice images, which can solve the conflict between simplifying image for watershed based on H-minima and preserving the object edge. Iterative bilateral filtering is first used to preprocess the image and it can preserve the edge well. Watershed segmentation based on H-minima can then used in the filtered image. By incorporating bilateral filtering with watershed transform, oversegmentation is significantly reduced and boundaries of major objects are precisely located. The proposed method achieves efficient and accurate performance in the auto-segmentation of natural images and SAR sea ice images, and the new method can also identify filament structures such as leads.
Date of Conference: 25-27 June 2008