We develop the clustering approach by fast-search-and-find of density peaks which use the kernel function for mapping the input sample objects to the high dimensional feature space and amplifying the original sample objects¡¯ characteristics, thereby researching the local density of the object attribute based on similarity measure. In order to identify our method¡¯s effectiveness, we compare our method to other popular clustering algorithms. From the result analysis, we could conclude that our method have a better result.