Prior Label Based Sub-Markov Random Walk For Efficient Image Segmentation

Abstract

 Segmentation is the first step in object identification in any image. It can also be used to compress different areas or different segments of an image, at different compression qualities. So, for the segmentation of an image we have developed a novel technique known as sub-Markov random walk (subRW) algorithm with label prior for seeded image segmentation. This is similar to the traditional random walk with auxiliary nodes added in it. Under these auxiliary nodes consideration we have given uniqueness of proposed work than existing systems. Our method is more efficient compared to previous work. The uniqueness will be nothing but adding or changing the auxiliary nodes in segmentation algorithm. We face segmentation problem in existing system if the image is having very thin and elongated parts. To solve this type of problem we implemented proposed work. Matlab simulation results proved that our proposed subRW is giving better results compare to all other existing RWalgorithms.