In this paper, a new approach is proposed for personal authentication using patterns generated on dorsal of finger. The texture pattern produced by the finger knuckle is highly unique and makes the surface a distinctive biometric identifier. Important part in knuckle matching is variation of number of features which come by in pattern form of texture features. In this thesis, the emphasis has been done on key point and texture features extraction. The key point features are extracted by SIFT features and the texture features are extracted by Gabor and GLCM features. For the SIFT and GLCM features matching process is done by hamming distance and for the Gabor features matching is done by correlation. The database of 40 different subjects has been acquired by touch less imaging by use of digital camera. The authentication system extracts features from the image and stores the template for later authentication. The experiment results are very promising for recognition of second minor finger knuckle pattern.