Based On The Head Of The Sub-pixel Edge Features Image Registration And Fusion Research

Multi-modality medical image fusion plays an important role in medical analysis and diagnosis. Multi-modality image registration is the basis and the difficulty of image fusion. As the basic issue of medical image fusion, medical image registration has very important meaning. It is the first step of image fusion, and its precision influences the effect of image fusion directly. The aim of medical images registration and fusion is to align the same or different modal images of the same or different research object in space location, and integrate different modal images into one image, then it can be available for disease diagnoses, treatment and evaluate after operation, so it is very important for theory and application research. The paper focuses on Multi-modality medical cerebral image registration and fusion, including image preprocessing, subpixel edge extraction, cerebral image registration and fusion algorithm.The aim and significance of the subject are discussed firstly. Medical image registration and fusion method, including classification principle, applied field and development, are reviewed then. At last, the paper analyses image preprocessing, subpixel edge extraction, registration and fusion methods in detail and validates them by experimentation.In preprocessing, because of the different data format, the data is transformed firstly. To process easily, the transformed data is normalized.Edge feature is the basis image feature and usually used for image registration. Edge extraction plays an important role in medical image registration based on edge. Now most edge extraction algorithms in medical image registration are pixel level. Low precision of the edge leads to low precision of registration. A new fast medical image subpixel edge extraction algorithm is put forward in the paper. After combining pixel-level operator with zernike moment operator and search algorithm, the results of edge extraction are better. The algorithm improves the precision of the edge, reduces the time of processing and establishes nicer foundation for thesubsequent workOn the basis of edge extraction, the paper uses SVD-ICP algorithm to align multi-modality medical cerebral image and validates the accuracy and validity of the method by experiment.At last, the paper uses threshold, local fit and wavelet transform modulus maximum methods to fuse respectively the multi-modality medical cerebral image. The result of experiment indicates that wavelet transform modulus maximum method is better. The fusion images are clear and give prominence to available information of different medical cerebral images better.A new experiment result evaluation method is put forward. The existing evaluation methods are low accurate or complicated. My method operates image matrix directly. It is intuitionistic, effective and has good effect.In a word, through thorough and systemic research for subpixel edge extraction, the registration based on feature edge and the fusion, an adaptive algorithm is proposed. Large numbers of images and data have been used to prove the accuracy and the innovation of my methods in this paper...