Abstract: Visual pattern recognition using face and vein images as input to identify someone is not a new issue in the filed of pattern recognition. However, to acquire a face image is totally different with a vein image. In addition, factors that affect the recognition rate among them are also different. This study describes simulations that were conducted on three kinds local descriptors, i.e., LBP, LDiP and LDNP to represent face and vein images for identification. The classification is using Chi-square histogram matching as a dissimilarity measure. From the simulations conducted it was shown that the local-feature descriptor was not properly applied directly for vein images.