Study On Registration Optimization For High-Resolution Light Microscope Images

Three dimensional reconstruction of biological tissue slice image plays an important role in studying the organ and tissue structure. But in the image acquisition procedure, sequential images are not aligned,such as translation and rotation. These images must be registrated before using. But efficient method for high resolution images registration is rarely.In this paper, a new method for the registration of high resolution sequential biological tissue images is proposed. Firstly, the conception and basic methods are introduced, and then the method for high resolution image registration is present according to the property of the images. The mean square difference of gray scale is selected as the similarity critical. LM algorithm and B spline interpolate are used. Pyramid algorithm is used to be optimized the time computation and avoiding the local extremum error.Result shows that this method is robust for slice image registration in avoiding local extremum and keeping convergence. The pyramid algorithm reduces processing time and improves the efficiency. This method can be broadly used in image registration, which is useful in morphology study.