Research On Medical Image Segmentation And Reconstruction Algorithm Based On Multiple Information Combination

Digital image processing refers to the use of computer technology to extract some information in the image that is suitable for observation,understanding and recognition analysis.Image segmentation,as an important branch of digital image processing domain,is mainly based on certain image features and extracts areas of interest from background or other pseudo targets.It has a predecessor role in image analysis and pattern recognition.Therefore,the application of image segmentation technology in medical images can provide a reliable basis for clinical diagnosis and pathology research,and assist doctors in making more accurate diagnosis.Among current image segmentation methods,the level set method is widely used because it uses dynamic evolution curves to handle its free topological changes.However,the problems of uneven gray levels and blurred borders between organs and tissues in some medical images have resulted in inaccurate segmentation results of the level set method,making accurate and effective distortion-free segmentation algorithms a hotspot in the field of medical image processing.In addition,in order to better display the complete shape of organs and tissues,a moving cube algorithm is used to reconstruct the segmented two-dimensional image into a three-dimensional stereo image.However,the existing moving cube algorithms have problems such as ambiguity and high computational complexity.In view of the above defects,this paper selects medical images with blurred borders and uneven grayscales as research objects,and uses the level set method based on the boundary information,area information,and feature point information of the medical image to segment the image.Then the segmented image is used The improved moving cube algorithm reconstructs two-dimensional images into three-dimensional stereo images.The main work and contributions are as follows:1.Aiming at the problems of blurred boundary and gray unevenness of medical images,this paper proposes a medical image segmentation algorithm based on information-integrated level set.By establishing the coefficient relationship between area information and boundary information,reducing the setting of coefficients and increasing the algorithm The scope of application,combined with the distribution information of the feature points of the image,uses the KL(Kullback-Leibler)divergence to maximize the distance difference between regions,accelerates the evolution curve to the target contour,and reduces the time complexity of the convolution operation in the algorithm,While ensuring the performance of medical image segmentation,it also improves the segmentation efficiency of the level set algorithm.At the same time,in order to solve the problem of improper selection of extreme points in gray uneven images,an evolution method based on momentum and stochastic optimization is proposed.By combining the previous gradient information,the non-convexity function is reduced to make the evolution curve uneven in gray.The possibility of falling into a local minimum under the environment,realizing the correction of the moving direction of the evolution curve,and improving the segmentation performance of the level set algorithm in medical images with uneven grayscale.2.Aiming at the problem that the ambiguous surface of the traditional moving cube will cause the void effect and high computational complexity of the reconstruction model,based on the original hyperbolic method of judging the ambiguous surface,an improved moving cube algorithm is proposed.The intersection point between the hyperbolic isosurface function and the ambiguous surface edge is replaced by the intersection point between the hyperbolic asymptote and the ambiguous surface edge,which improves the calculation efficiency of the algorithm.In the reconstruction process,because linear interpolation increases the time and computational complexity of the algorithm,linear interpolation is improved to center point interpolation,which reduces the reconstruction time and improves the three-dimensionality of the reconstructed medical image.