Improved Stochastic Random Walker Segmentation based on Gaussian Filtering

Abstract

Image segmentation is the process to capture the object from the background and it is a difficult task when a vision of the object is in stochastic region. Here introduce in this paper extension of stochastic random walker segmentation method. In stochastic random walker segmentation, a weighted graph is built from the image, where the each pixel considered as a node and edge weights depend on the image gradient between the pixels. For given seed regions, the probability are evaluated for a stochastic random walk on this graph starting at a pixel to end in one of the seed regions. The problem associated with existing method is the number of random variable (gray-level value in random order) in stochastic images. These random variables increase the graph sizes of stochastic images. If the graph size will increase, consequently execution time would also increase. To overcome these problems, the proposed "Improved stochastic random walker segmentation based on Gaussian filtering" for stochastic image segmentation. In proposed method Gaussian filter has been used for the removal of uncertain gray level and which in turn reduce the noise level and the resultant graph size of corresponding stochastic image, then apply stochastic random walker segmentation method which may help to decrease the execution time of the segmentation process.