Research And Implementation Of Medical Image Segmentation Based On Improved Level Set Algorithm

Medical images have complex diversity on account of many factors.The original medical image usually contains many problems,such as uneven grayscale,weak edge,high noise,multi-target,etc.Traditional image segmentation algorithms are difficult to segment these images effectively.The level set(LSM)has attracted the attention of many researchers due to its advantages of low algorithm complexity and easy fitting of other algorithms.The level set algorithm has been broadly used when we need to segment images.However,in the process of complex medical image processing,a series of problems will occur in the horizontal set algorithm such as poor stability,poor robustness and large amount of calculation.In this paper,aiming at the different problems of medical image,the level set method is improved and fitted.The main research contents and results are as follows:1.It introduces the mathematical knowledge which are used in the level set segmentation algorithm,including the concept of partial differential equation,solution of partial differential equation,definition of variation,euler-lagrange equation,gradient descent flow and other knowledge.The research on the evolution of curve is done.This is the basic theory for level sets.Through studying the basic knowledge of level set,the basic model of level set is further studied.2.Aiming at the features of high noise,multi-target,and difficult segmentation of medical images,a MR image segmentation model based on wavelet de-noising and double level set is proposed.For the problem of large noise artifacts in medical images,wavelet transform is introduced to remove noise and the image is preprocessed.The multi-phase CV model is able to divide images into many targets,but the disadvantage of this algorithm is that the segmentation speed is slow.The energy penalty term is added to the energy function of the DCV model,and the traditional DCV model is improved to segment the image after wavelet de-noising.Finally,the segmentation of high noise and multi-target images is realized.3.In view of the noise and gray scale inequality in medical images,an image segmentation algorithm based on morphology and improved level set is proposed.The image is processed by using the knowledge of morphological reconstruction to reduce the image noise while maintaining the edge profile information of the target.The information of offset field is added to the energy function to solve the problem of image gray level inequality.By using the improved model,the preprocessed images can be divided into different gray-scale medical images.4.A medical image processing system has been designed,which software environment can allow developers to compile algorithms in M files.And the users also can improve the algorithm and change interfaces in M files.At the same time,it also allows most users to operate directly on the GUI interface,and the Human-Computer interaction is very good.