A face retrieval approach is proposed based on the re-sampling of interested region to deal with the problem that the distinguish ability of detection and retrieval is different for different parts of face. The approach inherits the advantage that the feature extraction after blocking the image can remain local information

and just re-samples important parts of human face by moving the sampling block in a certain sampling step. Thus important regions are sampled densely and meanwhile the blocks are overlapped in a specified distance. The approach not only avoids losing the judgment of improper classification in important regions caused by the uniform block

but also reduces complexity and enhances the anti-interference ability. The approach realizes the weighting of important region in face retrieval

and its performance is dramatically enhanced under the circumstance of little increase of computational complexity

compared with the existing uniformed blocking method. A large number of comparative experiments verify the conclusion.