Accelerated Q-learning approach for minutiae extraction in fingerprint image

Fingerprint recognition is a physiological biometric technique. It is most dependable as compare to other biometric technique. Fingerprint recognition involves preprocessing, minutiae extraction and post processing stages. In conventional approaches preprocessing stage include image processing steps to reduce noise. Image processing steps are extremely sensible against noise. A Q-learning approach used for minutiae extraction generates insensitiveness against noise but it also gives success for wrong ridge path which expend processing time. In this paper we have proposed accelerated Q-learning approach for minutiae extraction which calculate Q-value for both success and fail state. In proposed method we follows ridges if it gets fail state it leaves that ridge path otherwise it will continue to follow that ridge and calculate Q-value for success state. Proposed method reduces processing time and also improves efficiency against noise. Keywordfingerprint images, minutiae extraction, ridge endings, ridge bifurcation, fingerprints recognition.