Study On Medical Image Detection And Registration In Single - Phase Phase Consistency

CT, MRI, SPECT and ultrasound medical image contains its own emphasis on the characteristics of information. The image of a single mode is often not fully and accurately reflect the appearance of tumor type, etc., so the medical images of a variety of models for effective treatment, through the automatic registration of contrast can provide an important basis for medical diagnosis and treatment. But the medical image has low contrast, speckle and edge fuzzy feature extraction, clinical diagnosis information has a certain degree of difficulty in. The key technology of image registration is the accurate feature information extraction based on.Automated image analysis must first obtain interest characteristics more significant by the image conversion, and spatial gray method or contrast intensity based on medical image detection effect. The phase consistency method based on frequency domain image by Fourier transform, calculates the similarity between pixels in the phase frequency, take the highest similarity point of interest points, the ideal detection method, but the large amount of calculation. The monogenic signal theory of one-dimensional analytic signal is extended to the two-dimensional analytic signal through Riesz transform, can effectively simplify the calculation. In this paper, combining the two, detection algorithm of medical image points of interest monogenic phase consistency and edge is given based on the structure of the filter, monogenic phase consistency in low brightness and contrast conditions, accurate detection of image features.Mutual information technology for image registration based on gray, but all the pixels within the image traversal calculation. In this paper, the characteristics of monogenic phase coherence filter to detect the regional mutual information image registration technology, the extracted interest points and edges as the processing object, to calculate the registration in accordance with the mutual information criterion.Based on the cross section of the brain, mediastinal, liver and other medical image processing simulation, the experimental results show that:compared with the Canny operator, Laplace operator and other conventional methods based on spatial domain, monogenic phase congruency algorithm can be more rapid, accurate detection of interest points and edges, for the follow-up of the regional medical image registration mutual information algorithm based on the provided good precondition.