Applied Research On Medical Image Registration Algothim Based On Feature Region

In recent years, the medical image registration is one of the most popular field in the medical image processing. It attracted a vast number of scientific and technological worker study in this field. And laying a good foundation for the image application of image fusion and image mosaic. This paper based on the full understanding of the traditional image-processing methods and the previous work, systematic study the medical image registration mesthod based on the image edge feature point. And make a certain degress of improvement and innovation. Through a detail theoretical analysis and experimental verification. The main research contents are followed.Firstly, in order to solve the conflict between noise reduction and detection of image edge details, the wavelet transform with the characteristics of multi-scale was used. The image signal was broken into number of components on different scales and made the image multi-analysis. The relation between Lipschitz Regularity and Wavelet Modulus Maxima were used. The medical image edge detection algorithm based on wavelet multi-scale was proposed.Secondly, in order to overcome the Harris corner detection algorithm for the disadvantage of large calculation and inaccurate extraction, the multi-scale method was introduced to the process of extract image feature points and the original algorithms was made improved. A new Harris corner detection algorithm based on multi-scale was proposed .Thirdly, accurate set of match corner points pairs was made in the two registrating image based on the detected corner points. Three feature regions were selected based on part match feature corner pairs. The image registration was made based on similarity measure of this feature region the alignment degree.Finally, the experimental medical image be taken from Beijing Tiantan medical biotechnology company and simulate the arithmetics above in Visual C++. From the registration on the success rate, the improved arithmetics has increased the registration on the speed and accuracy. Compared with the arithmetics of MI and basically reached its registration results.