Image Segmentation Method Based On Watershed And Contourlet Transform

Image segmentation plays a very important role in digital image processing. The watershed transform is one of the commonly-used image segmentation methods. For rudimentary watershed produces excessive over-segmentation, a marker-driven watershed is usually employed to deal with the problem. A novel image segmentation method based on the nonsubsampled contourlet transform and watershed transform is proposed, which efficiently combines the nonsubsampled contourlet transform and watershed transform in mathematical morphology. Firstly, we modify the image gradient by the nonsubsampled contourlet transform. Secondly, a marker-driven watershed transform is used on the modified gradient image. The experimental results show that the proposed method is feasible and can give better image segmentation. It does not only avoid the over-segmentation, but also preserves the edge information. At last, according to the segmentation result,we propose a region extraction method via the density of the gradient image. This method can further judge the types (edge, smooth or texture region) of the segmented sub-regions.