Research on SVM sea ice classification based on texture features

The classification of sea ice is one of the most important applications in the field of remote sensing monitoring, and its accuracy is of great significance in assessing the ice conditions, ensuring the safety of navigation and opening up the Arctic channel. In order to solve the sea ice classification problems, this paper proposed an improved SAR sea ice classification method, which used Sentinel-1 data and texture feature analysis. In this method, the gray level co-occurrence matrix (GLCM) was used to extract the eigenvalue, and the suitable of texture features for sea ice classification was obtained, then we used support vector machine to carried out sea ice classification. The experimental results showed that the proposed method can recognize three types of ice, which are first year ice, multiyear ice and open water. Compared with the traditional methods of Neural Net and Maximum Likelihood, it is feasible to use SVM classification method and texture feature to monitor sea ice type. It also showed that multi-feature is helpful to improve the classification accuracy of SAR image, which verifies the effectiveness of this method and provides a new idea for sea ice classification.