Off-line Signature Verification by Matching With a 3-D Reference Knowledge Image: from Research to Actual Application

This paper introduces a method for off-line verification of signatures. The method can verify signatures and detect skilled forgeries with outstanding performance. It is based on matching the questioned signature with a 3-D Reference Knowledge Image (RKI) using Ammar Matching Technique (AMT). The AMT which was developed and modified through the years (1989-2010) is introduced in detail, and the 3-D RKI derived from the methodology of the Forensic Document Examiner (FDE) of working is elaborated. The features extracted using the new knowledge representation method and the AMT have been found to be highly effective in signature verification, and distinctly outperform the classical features like slants, baseline, and contour based ones on the data used. Experimental results of using RKIs built from binary, High Pressure, and thinned images for feature extraction and verification are also presented and discussed.