Abstract: Based on the principle of fuzzy pattern recognition and the characteristics of kernelbased method, an algorithm of kernelbased fuzzy clustering is put forward. Based on the principle of fuzzy pattern recognition and the characteristics of kernel method, the objective function of fuzzy Cmeans is substituted with a kernel objective function and the Gaussian kernel function is used in a kernelbased fuzzy clustering. The approach of kernelbased fuzzy clustering is used in the classification of remote sensing image, and the result shows that it can effectively improve the classification accuracy of remote sensing images compared with the traditional fuzzy Cmeans clustering. The approach presented in this paper can develop the application range of fuzzy pattern recognition.