Fingerprint Enhancement Using Wavelet Transformation and Differential Support Vector Machine

The security measurement relies on knowledge-based methods such as token (swipe cards (credit cards) and passports to control access to physical and fundamental spaces), passwords (Net banking, ATM Cards). Though ever-present, such techniques are not secured. Passwords like badges and cards might be shared or stolen. Furthermore, they cannot not differentiate between the authorized person and an unauthorized user. Hence, biometric traits like face, retina, palm and fingerprints are used for authorization. The fingerprint is the most famous biometric traits used for security and law-enforcement. Our research work, introduces a novel method for fingerprint image enhancement using DWT, PDE and SVM (matching and classification). Discrete Wavelet transformation method is used to filter the latent fingerprint images in LL, HL, LH and HH bands. The fingerprint minutiae methods are extracted by screening the local point of each ridge pixel in the image using a 3×3 window. PDE algorithm fetches the improved features through minutiae method. The SVM classify the unique features in two phases: (i) Training Section and (ii) Testing Section). The proposed system has given better performance with GAR 0.8%, Peak Signal to Noise Ratio value is 54% and FAR value 1 to 5 per cent. PDR (Partial Difference Rate) and Gaussian filter is used for the improvement phase in Partial and classification fingerprint verification.