ABSTRAKSI: Kata Kunci : ABSTRACT: Identification techniques have been widely applied for recognition of human identity, many of which are in the form of a password security system or identity card. Where the introduction of human identity through fingerprint, iris, face and DNA, has been done and researched accuration value. While the introduction of human identity through newly developed finger knuckles. In this final has been researched, designed and analyzed a system that can identify people using finger knuckle. Imagery used is finger left hand, a finger, sweet and center. Using digital image processing and extraction feature using 2D Gabor wavelet filter and ANN classification process using Learning Vector Quantization (LVQ). the selection process vector feature value used additional techniques such as Feature Wrapper Subset Selection. At the end of the task previously used methods PCA feature extraction and classification using the K-NN values obtained perfect accuracy of 100%.However, in this thesis by using 2D Gabor wavelet filter and classification method Learning Vector Quantization (LVQ) results obtained from the value of accuracy in identifying human identity value generated the highest accuracy of 77.5% on the training images and 41.67% on the test images to 24 vector characteristics, while using 15 vector characteristics of the resulting value of the highest accuracy of 38.33% on test images. It can be concluded that by using the method of feature extraction of 2D Gabor wavelet filter is not very reliable and appropriate for the type of media biometric finger knucklesKeyword: Identify, finger knuckles, 2D Gabor wavelet filter, LVQ, digital image processing, wrapper