Off-Line Skilled Forgery Detection Using Stroke and Sub-Stroke Properties

Research has been active in thefield offorgeiy detection, but relatively little work has been done on the detection of skilled forgeries. In thispaper, we present an algorithmfor detecting skilledforgeries based on a local correspondence between a questioned signature and a model obtained a priori. Writer-dependentproperties are measured at the substroke level and a cost function is trainedfor each writer. Whena candidate signature ispresented, the samefeatures are extracted and matched against the model. Wepresent a description of thefeatures and experimental results.

• Type: Conference
• Year: 2000
• Where: ICPR
• Authors: Jinhong Katherine Guo, David S. Doermann, Azriel Rosenfeld
• Keywords: Candidate Signature, Computer Vision, ICPR 2000, Skilled Forgeries, Thefield Offorgeiy Detection