HaFT: A Handwritten Farsi Text Database

Standard databases provide for evaluation andcomparison of various pattern recognition techniques bydifferent researchers; thus they are essential for the advance ofresearch. There are different handwritten databases in variouslanguages, but there is not a large standard database ofhandwritten text for the evaluation of different algorithms forwriter identification and verification in Farsi. This paperintroduces a large handwritten Farsi text database called HaFT.The database contains 1800 gray scale images of unconstrainedtext written by 600 writers. Each participant gave three separateeight-line samples of his handwriting, each of which was writtenat a different time on a separate sheet. HaFT is presented inseveral versions each including different lengths of text and usingidentical or different writing instruments. A new measure, calledCVM, is defined which effectively reflects the size of handwritingand thus the content volume of a given text image. This databaseis designed for training and testing Farsi writer identification andverification using handwritten text. In addition, the database canalso be used in training and testing handwritten Farsi textsegmentation and recognition algorithms. HaFT is available forresearch use

کلیدواژه ها:

handwriting database ، Farsi ، HaFT ، writer identification ، writer verification ، content volume measure ، CVM