Image Registration Method Based On Wavelet Theory Research

Image registration refers to geometric transformation of two or more images for matching them. The Images are got from the same scene in different views or under different conditions using the same or different sensors.The purpose of image registration is to eliminate the differences between two or more images, to obtain more comprehensive and more accurate information of image description. Image registration is one of the classical problems in digital image processing. It has been widely used in identification of military target, diagnosis of tumor, fusion of remote sensed images,etc. In recent years, the theory of wavelet transform has attracted more and more attention of scholars. To achieve higher efficiency and precision of registration, scholars try to combine feature-based image registration method with wavelet transform. The combination of wavelet transform, the corner feature detection or various kinds of distance measurement method is effective.And a lot of experiment have confirmed the feasibility of these combination algorithm. This paper first outlines the development of image registration technology and the present research situation. Summarizes the existing main methods of image registration, make key research on the basic flow of feature-based image registration method. Then introduces the structure of the wavelet transform, which educe the structure and realization of the multi-wavelet. Make key research on the structure and character of the GHM multi-wavelet, describe the application of GHM multi-wavelet in image processing. Thirdly, describe a method of feature extraction which is based on Mean-shift segmentation algorithm. Introduce features invariant descriptor, and put forward a kind of image registration method based on feature similarity. Finally realize the algorithm with Matlab software. According to the experimental results, analyze the comprehensive performance of the proposed algorithm in this paper.