Biometric is a method to iden the specific feature based on part Nowadays, biometric is widely used performance to identify human. The pa used in this research were manually coll of 110 people with 10 images for each p recognition system consist of : region extraction using Competitive Hand (CHVD) method, feature extraction pro Binary Pattern (LBP) method and clas using Probabilistic Neural Network research we proposed area segmentati LBP that consist of overlap and non-ove using segmentation the feature will beca and distinctive. In 60 people (50 known the best system accuracy achieve was 9 110 people (100 known and 10 unknown) achieve was 91.91%. The best performan with four overlap segment schemas, 8 che smoothing parameter set at 0.03.