Core-point Locating And Fitting Of Log-spirals In Typhoon Satellite Images

Log-spiral (or equiangular spiral) is a special spiral curve which often appears in nature. Because it has good geometry, mechanical and aesthetic properties, so it has been widely applied in various fields, including computational geometry, industrial modeling, astronomy, meteorology, geography and animal research.This thesis mainly discusses the problem of detecting, extraction, fitting of log-spiral in typhoon satellite images. For detecting and extracting the initial data, it converts the original image into a binary image by the Otsu threshold selecting method firstly. Then using the method of mathematical morphology and the feature of area extract the typhoon cloud department, finally, extracting the skeleton which represents the shape feature of spiral in images.In the processing of log-spiral fitting by least square method, it mainly considers the location of a center point and the interference from outliers. In order to improve the efficiency of the algorithm, we narrow the scope of the center point with the related properties of log-spiral when searching a center point in images. For the robustness of the algorithm we detect the outliers using the method of VfM (variance-from-median) and remove them by iteration to reduce the interference from outliers.