Handwriting identification from the perspective of optimization model

Handwriting is a skillful graphical shapes accomplished by human on a surface paper, wood, etc. Analyzing differences and similarities between writers in order to identify the authorship of handwritten document is called writer identification. While invariant features are the core stone to classify the writers, the importance of a specific feature has not been investigated. This study aimed to examine feature importance in writer identification using Binary Particle Swarm Optimization (BPSO) algorithm. Off-line text-dependent words from IAM database are used. Moment and statistical features are extracted to represent the handwritten words. A significance influence of the feature weight is being showed by the experiments results.