Multi-Layer Framed Offline Signature Recognition Algorithm
Wiraj Udara Wickramaarachchi, Shanmuganathan Vasanthapriyan · Journal of Image and Graphics · 2015
This research experiment ventures to find a solution for automating the framed signature recognition. Here signatures are made on a given frame using a ballpoint pen with a tip size of 0.5. Instead of direct neural networks based algorithm implementation, the extracted non-scale variant and scale variant features are used in a support vector machine in signature recognition algorithm. The outcome of the research appears as a GUI. The final outcome was 100% random signature isolation with over 88% trained forgery rejection. If not for 4 vulnerable signatures, this rate goes over 96% however the research carried out with worst environmental conditions and with least number of features. Thus, the results can be definitely improvable with modifications.