Abstract

Info

Title:: A New Segmentation Algorithm Based on Snake Model for Brain CT Image

Author(s):: XU Yan1; 2; HU Shunbo1; WANG Jifeng1; DU Yuyue2; 1. College of Information Science and Engineering, Linyi University, Linyi 276000, China 2. College of Computer Science and Engineering, Shandong University of Science and Technology, Qingdao 266590, China

Abstract:: Aiming at the disadvantages of traditional Snake model-based image segmentation algorithm, such as small force field capture range, sensitivity to initial contour selection and the difficulty of converging contour curves to small deep concave boundaries, a novel Snake model-based segmentation algorithm for brain CT image is proposed. Firstly, Canny edge detection operator is used to detect the edge of the image, and the obtained edge detection image is superimposed onto the original image. Subsequently, Snake model and gradient vector flow (GVF) Snake model are applied to segment the superimposed image. The experimental results show that the proposed algorithm overcomes the missed segmentation by traditional Snake model and GVF Snake model due to unclear edge contours, and prevents the over-segmentation by GVF Snake model due to GVF force field interaction. Meanwhile, the proposed algorithm can also converge the contour line to to a small deep concave boundary, improve the positioning accuracy and has a better segmentation result.