When using the existing methods to carry out the enhancement of the low resolution palmprint images, Scale invariance feature transformation (SIFT) keypoints used for identification can not be efficiently yield. To solve this problem, this paper proposes a promising local entropy-based Unsharpen Masking algorithm for SIFT feature extraction of the images described above. This method can adaptively improve the contrast of the images to extract SIFT features from palmprint while it does not distort the characteristic of the palmprint texture. The scheme introduces local entropy of the images as the enhancement coefficients. The experimental results show the proposed method can successfully extracts as many as possible SIFT keypoints from the enhanced palmprint images, in contrast to the failure by traditional Unsharp Maskingbased schemes.