Comparison of Online Writer Identification Systems for Limited Data

Writer identification systems attempt to determine the identity of the author, given a piece of handwritten text by comparing it against samples of known authorship. They are categorised into Online and Offline on the basis of how data has been captured. Another classification of writer identification systems is text independent and text dependent systems based on the textual content provided. Existing methods in literature for online text independent writer identification make use of a training set that consists of numerous paragraphs of handwritten data. For a practical scenario, this may not be always available. This paper aims to compare the performances of existing text-independent online writer identification systems for the scenario of limited training data and investigate the shortcomings of these systems.