Sift technique on extraction of fingerprint features

Abstract

This project has a final goal of implementing the Scale-invariant feature transform (or SIFT) algorithm towards fingerprint features extraction. The algorithms comprise of scale space construction, keypoint localization, orientation assignment and keypoint descriptor. The scale space construction is using the DOG to detect stable key points and performing neighborhood comparison to detect the scale space extrema. Next the keypoint localization algorithm will be using the Taylor Expansion theory to reject the unstable keypoint which is low contrast. Subsequently, the orientation also will be assigned to each keypoint location based on local image gradient directions. Lastly, the keypoint descriptor is used to compute descriptor vectors which is highly distinctive. After implementing the SIFT algorithms, it is used to validate against all sort of common invariance and the outcome results are showing good accuracy. Conclusion, SIFT finds accurate features against those common invariances, such as scale invariance, rotation and illumination.