Tropical cyclones develop into typhoon cloud system has several changes

The typhoon is a cyclonic disturbance on the tropical ocean surface developed into a low-pressure system, is the strongest of the tropical cyclones. It is one of the disastrous weather that seriously affects human production and life, and our country is one of the most serious typhoon disasters in the world. In order to avoid the disasters caused by typhoons to human beings, the early forecasting of typhoons is of great significance. Moreover, the earlier the forecast is made, the less damage will be caused. Therefore, digital image processing is applied to segment the typhoon cloud system from various cloud systems during the formation period of the typhoon, and to extract and localize its features. However, from the current information, the traditional method of manual forecasting is still the main method for forecasting typhoons, while automatic forecasting of typhoons in the formation period is still in the research and exploration stage. Since the goodness of typhoon center localization depends on the segmentation and feature extraction of typhoon cloud system. Therefore, the segmentation and feature extraction of the typhoon cloud system has high research value and extremely important practical significance.? The main work involves the following aspects:?1. Preprocessing of the original cloud image. Firstly, the color image is converted into a gray-scale image, and then an appropriate threshold is selected to filter out a part of the cloud map that does not belong to the typhoon cloud system, and then a method combining recursive median filtering and mathematical morphology is proposed to filter the cloud map in order to eliminate the interference of noise. The preprocessing work greatly simplifies the subsequent work. ?2. Segmentation of cloud maps. According to the typhoon cloud system has the characteristics of relatively high gray value, large area and small eccentricity, the cloud map is segmented by the gray threshold method, area segmentation and eccentricity segmentation respectively to segment the typhoon cloud system from various cloud systems. ?3. Extraction of spiral cloud bands. Firstly, the feature information of the typhoon cloud system is recovered, and then the spiral cloud bands of the typhoon cloud system are extracted by the difference method, and the extracted spiral cloud bands are skeletonized and pruned by the mathematical morphology method, so as to get the local spirals of the spiral cloud bands. ?4. According to the features of the extracted local spirals, the least squares method is used to automatically locate the center of the typhoon.