Research On MRI Segmentation Method Based On Bias Estimation

Image segmentation is the process that dividing the image into a number of areas based on the gray,color,shape and other features.The image segmentation is one of the important aspects of image processing.Medical images provide doctors with rich information about tissue and organs.With the widespread application of medical image technology in clinical diagnosis and treatment,medical images have become an important tool for disease diagnosis.However,the segmentation of images by artificial methods can not meet the requirements of fast and accurate segmentation,and efficient and automated image segmentation methods need to be applied to medical images.But,due to the special imaging principle of medical images,the general image segmentation technology can not segment the medical images very well.So that it is required to put forward more appropriate segmentation methods for medical images.In order to solve the.problem of nonuniform intensity of MR images inmedical images,this paper adopts the methods of multiplication factor optimization and intensity averaging to segment MR images.The image is expressed by the product of the bias term and the true term.Through the process of minimizing the energy equation,the parameters of the bias term and the true term can be calculated,which are the estimation of the image bias field and the segmentation result.For the nonuniform intensity caused by bias magnetic field,we can obtain the real image by estimating the bias term.In order to improve the convergence rate of the energy equation,a constraint term is added to the energy equation.The another way we used is averaging the intensity of the local region in image and take it into the multiplication equation of the image.At the same time,the image segmentation results are obtained by the level set method in combination with the global information of the image.The process of intensity averaging relieves the blurring of the boundary in the image,enhances the intensity difference between adjacent regions,and improves the segmentation accuracy.By using the complementarity between local information and global information of the image,the segmentation effect is improved,and the possibility of the algorithm falling into local minimum is reduced.Experimental results show that the proposed methods have good segmentation effects for different degrees nonuniform of MR images.For the blurred images,the segmentation accuracy of the proposed method is also high,so it has good robustness.