Fingerprint Recognition Based on Region of Interest (ROI) Extraction from Fingerprint Images

Fingerprint authentication is one of the most developed and well-known biometrics because of its better accountability, better security and uniqueness. Common fingerprint recognition techniques which are based on complete fingerprint image do not perform well when fingerprint image is noisy or incomplete/partial. To overcome this drawback, we utilize part of the image which present in fingerprint image whether it’s full or partial. Some portion of image which conveys meaningful information and consistent in every image refer as Region of Interest (ROI). We propose a mechanism which extracts ROI by removing inconsistent boundary region. Then resize image that called ROI of the image. Our proposed method performs effectively and accurately even though fingerprint image is noisy, incomplete/partial and not contains any reference point. By working on subset of image, the processing time is reduced. Thus, the fingerprint recognition becomes fast. Broad experiments over Fingerprint Verification Competition (FVC) database demonstrates the advantage of the proposed approach compared with different methods in literature.