A Method of Image Mosaicing Using Log polar Coordinate Mapping
基于对数极坐标映射的图像拼接方法

Image mosaicing has many applications, such as image based rendering, video retrieval ,scene matching, etc. Aiming at images with rotation and scale, a method of image mosaicing based on log polar coordinate mapping is put forward to obtain scene representation of wide view field .In this method, firstly each image is transformed from Cartesian space to log polar coordinate space, which causes scale and rotation in Cartesian space to be manifested translation in log polar coordinate space, and the translation between the pair of input images is calculated ,then phase correlation method is used to estimate rotation angle and scale factor between the pair of input images, and then these motion parameters are refined by nonlinear least square iterative optimization algorithm, lastly image blending is imposed to implement image mosaicing. The experiments show that it is an available method of image mosaicing.