Parallel Segmentation Approach of the CT Image Based on SVM
Peidong Wang · Harbin Ligong Daxue xuebao · 2013
In this paper,a new parallel segmentation approach using regional growth with support vector machine is proposed.The conventional regional growth is a difficulty to determine the feed points automatically,and a solo support vector machine is resultful in segmentation,but the speed is slow.In order to solve these problems,an image segmentation method combining support vector machine with regional growth was proposed.Firstly,training the support vector machine classification;then the trained classification is used to search seed points,and a curvature flow filter is used to reduce the noise and get a result with sharp and smoothing boundaries.Finally,regional growing with a simple but efficient threshold method is used.The experiment is performed on a parallel environment based on torque.Its result shows that the algorithm is feasible and work better and more faster than conventional algorithm.