Determining the Personal identity based on Handwriting as a Biometric identification
Mimoza Klekovska, Cveta Martinovska Bande · Goce Delchev University Repository (Goce Delčev University of Štip) · 2014
This paper describes methods for off-line identification of the writer based on handwriting features. Different methods for extracting and combining features are reported in the literature for pattern recognition purposes. Many aspects have influence over the writer identification such as: symmetry, slant angle, percent of black and white pixels, height/width ratio of the letters, direction of the base line, the position of the horizontal and vertical lines in the segments, histograms, contour profiles, spots, etc. The method creates a feature-vector associated with a writing manner of an individual and computes the correlation measure to express the similarity with the previously stored handwritten samples of the Cyrillic letters. The presented system is based on image processing and pattern recognition methods. The approach analyses the handwriting as an image-texture, it is content independent and uses feature set based on the global statistical, structural and topological characteristics. An experiment was performed to discover the most reliable features that contribute to the identification of the writer. Handwritten biometric identification is applicable in many areas such as: security systems, forensics, financial etc.