Segmentation Of Medical Image Processing And Image Reconstruction System

Medical images process and 3D reconstruction are the hotspots in the field of computer vision, they relate to the subjects of computer graphics,digital image processing,biomedicine engineering and so on. Medical images are very important in diagnostic, surgery planning and simulating, teaching in anatomy and medical simulate training. In order to dig out accurate and deep-seated information, modern technology is applied to the medical images.Medical image segmentation is an important part of visualization research. Meanwhile medical image segmentation is a hotspot in the field of computer visualization and a different science problem. The capacity of medical image algorithms which influence the precision of 3D reconstruction model is the foundation of 3D reconstruction.In the process of medical image transmit, the quality of image always descend. The preprocessing before image segmentation could increase the image quality, the usual methods are image filter,image emendation,image register and fusion. In this dissertation, at the foundation of compare with the 2D picture in the median filter, the mean filter and Lee filter's, use the adaptive window method for the medicine image filter, changes the size of window through the statistical property of the pixels which around it. According to this method, estimate the neighborhood of the central picture pixel. At the same time, this method has virtue of retention detail and noise elimination, Achieves the goal of filter effect which adaptive guaranteed the edge.The image preprocessing is the good foundation of image segmentation. In this dissertation, medical image segmentation algorithms are divided into based on edges' and based on regions'. Introduce classical algorithm Region Growth method in details, the result of which is influenced by the seeds and growth algorithm seriously. This article proposed a segmentation method based on support vector machines (SVM) and the Region Growth method. After this algorithm segmentation, the same aim possibly was divided into many modules. These segmentation blocks must be fused, according to the pattern recognition theory, if a group of training sample can separate by most super plane; the expectation of the upper boundary of the classification error rates is the proportion of the average support vector occupies overall trains' sets. This method reduce the subjective factor, enables the uniqueness of segmentation result.The surface rendering algorithms and volume rendering algorithms both have advantages. Though comparing the different algorithms, the surface rendering algorithms and volume rendering algorithms could use in different condition.A common platform medical images process and 3D reconstruction system is designed and developed, which is based on research some of algorithms of medical image preprocessing,medical image segmentation and 3D reconstruction.