Identification and Categorization of Rain-Contaminated Regions in X-Band Marine Radar Images

A method based on marine radar images in the X-band and the rainfall transmission (RT) feature is proposed to identify and categorize regions with varying rainfall intensities. The RT feature of each pixel is calculated based on a single image haze removal model, considering that the backscattter intensity varied with the regions of rain-free, light-rain and heavy-rain. A support vector machine (SVM) classification model is trained to segment different rainfall regions. The datasets used for training and testing were collected from a coastal radar located on Pingtan island, Fujian Province. Experiment results demonstrate that RT feature effectively segments radar images into distinct rainfall intensity regions, achieving higher accuracy with lower computational cost compared to the method using intensity histogram feature.