A novel low contrast IR image enhancement method based on multi-scale mirror-extended curvelet transform was proposed to solve the problems of the palm vein IR images, such as low SNR (Signal/Noise) and low grayscale contrast as the results of incorrect feature extraction of palm vein. Based on the analysis of the strong relationship between multi-scale curvelet coefficients and different scales of detailed vein features, coefficients of high frequency subbands, where most of the noises and few features located, were set zero. The coefficients of middle frequency subbands, where most of the features concentrated, were nonlinearly enhanced during the denoising process. The coefficients of low frequency subbands who determined the global grayscale contrast were stretched. Experimental results show that the proposed method can efficiently enhance the features of low contrast palm vein IR images with increased evaluation indexes such as SNR and Entropy. By this method the features of vein edges are better preserved and more smoothly emphasized than enhancement methods of biorthogonal wavelet and histogram equalization.