Region Growing and Level Set Synergetic Algorithms for Image Segmentation
Liu Qingling, Lin Ye, Faizan Ali Siddiqui · 2020
The application of segmentation technology in CT images can not only retain the clear characteristics of CT images, but also have the advantages of high resolution, clear anatomical relationship and clear pathological development. Level set segmentation method combines region and boundary information to solve the problems on complex texture and blurred edge of CT image. Therefore, this paper proposes an algorithm for regional growth and level set interaction for the problem of boundary sensitivity and lack of regional information in level set segmentation. In order to improve the efficiency and reduce the error of segmentation, this algorithm uses Gaussian filter to remove noise. Meaning while it adopts Ostu method to transform the image into a binary image. In addition, this algorithm provides an ideal growth environment for the seed point of the region growing algorithm correspondingly. The results of region growth are accomplished as the initial contour of the level set to modify the initial boundary. The boundary indication function introduces the region information; the level set segmentation algorithm achieves the purpose of accelerating in the pixel similar region and stopping at the target boundary as a result. Comparing the overall improved algorithms with the previous both region growing method and the level set method, the experiment results shows convincingly that the interaction between the level set segmentation and the region growing reduces the number of iterations. Under the condition of weak boundary and high noise, the boundary indicator function can accurately and completely segment the organ region and save the details of the internal organ completely, which improves the segmentation accuracy to a certain extent.