A Novel Medical Image Registration Algorithm Based on Contour Features and Multiwavelet Transform

摘要

The method of medical image registration based on image's contour features and multiwavelet transform is proposed in the paper. The method consists of two main procedures rough registration and accurate registration. In the rough registration procedure, Canny operator is selected to extract the images' edge, and the contour features are extracted based on the edge images. Then the Principal Axes method is used to get the translation and rotation information based on the images' contour features. Thus, the rough registration procedure can be achieved. In the accurate registration procedure, the couple medical images are decomposed by the simplified multiwavelet transform. And only the high frequency coefficients in the horizontal directions are selected as registration objects. Then the registration process starts from the coarse scale, and ends to the fine scale based on the images' high frequency coefficients and the rough registration information. At last, the two initial images are selected as registration objects to accomplish the accurate registration based on all the above registration information. Experiments show that it's an effective and accurate registration method of medical images. Furthermore, the results demonstrate the accurate registration under noisy environment too.