We present a method for writer identification that combines Forensic Handwriting Examination best practices with Pattern Recognition methodologies. The method is based upon a statistical characterization of the variability exhibited by a set of features that are meant to capture the distinctive aspects of document layout and handwriting. The features are quantitatively evaluated using a tool based on a model of handwriting generation and execution. The experimentation has been conducted on a database of handwritten documents produced by different writers using different writing modalities (spontaneous and copying). The experimental results confirm that the proposed method captures the distinctive aspects of handwriting and it is able to characterize the intra-writer and inter-writer variability and therefore to identify the writer of a questioned document in most cases.
|Titolo:||Combining FHE features with machine decision making for automatic writer identification|
|Data di pubblicazione:||2017|
|Appare nelle tipologie:||4.1.2 Proceedings con ISBN|