Offline text-independent writer identification using codebook and efficient code extraction methods
Image and Vision Computing
| Hi-index | 0.00 |
To identify a person using his/her handwriting; it is necessary to analyze a set of handwritings. Because of special styles of Persian handwritten; identifying Persian handwriting needs different approaches in compare with to other languages. This paper introduces a writer identification method to identify the writer of a Persian handwritten text. In the proposed method, the fuzzy approach is applied for the classification and clustering of handwritten texts. The proposed method consists of three main steps: the first step uses grapheme feature to select first candidates, the second one uses area features and fuzzy approach to restrict candidate domain; and finally third step uses gradient features to finalize selection. The experimental results were satisfactory and the accuracy of system was about 90% for 50 writers