Writer identification: A comparative study across three world major languages

- Accuracy of writer identification depends on window size manipulation.
- Accuracy of the writer identification deteriorates as database size increases.
- Data heterogeneity and human interpretability problems lead ambiguous accuracy.
- One size fit all writer identification model across various languages is proposed.

This paper presents a review on the state of the art in offline text-independent writer identification methods for three major languages, namely English, Chinese and Arabic, which were published in literatures from 2011 till 2016. For ease of discussions, we grouped the techniques into three categories: texture-, structure-, and allograph-based. Results are analysed, compared and tabulated along with datasets used for fair and just comparisons. It is observed that during that period, there are significant progresses achieved on English and Arabic; however, the growth on Chinese is rather slow and far from satisfactory in comparison to its wide usage. This is due to its complex writing structure. Meanwhile, issues on datasets used by previous studies are also highlighted because the size matter - accuracy of the writer identification deteriorates as database size increases.