In view of the problem that the traditional local binary pattern(LBP) and its extensions give the same eigenvalues and multiplication of feature dimension to the neighborhoods with different gray features, an algorithm based on the homogeneous k-means and high-dimensional local binary pattern is proposed. Firstly, the algorithm gets the sub-graph by cutting the original image, then extracts the high-dimensional local binary pattern characteristics of sub-graph and uses the homogeneous k-means to process the high-dimensional features by dimension reduction. Finally, the features of all the sub-graphs are cascaded to be analyzed. To verify the performance of the algorithm, the comparative experiments on the ORL face database, YALE face database and FERET face database are conducted, and the results show that the algorithm obviously improve the recognition rate of the LBP algorithm on the premise of ensuring that the feature dimension doesn't increase.