Farsi document image recognition system using word layout signature

Abstract

In this paper, a new representation of Farsi words is proposed to present the keyword spotting problems inFarsi document image retrieval. In this regard, we define a signature for each Farsi word based on the word connectedcomponent layout. The mentioned signature is shown as boxes, and then, by sketching vertical and horizontal lines, weconstruct a grid of each word to provide a new descriptor. One of the advantages of this method is that it can be usedfor both handwritten and machine-printed texts. Finally, to evaluate the performance of our system in comparison toother methods, a database that contains 19,582 printed Farsi words is examined, and after applying this approach, arecall rate of 98.1% and a precision rate of 94.3% are obtained.