Omnifont Persian OCR System Using Primitives

TL;DR

This paper presents an omnifont Persian OCR system that uses primitive structural features for recognition, achieving high accuracy across multiple fonts and sizes by integrating separation and recognition processes.

Contribution

The paper introduces a novel model-based OCR approach utilizing primitive elements for Persian script recognition, improving accuracy and efficiency over traditional methods.

Findings

Achieved 97.06% recognition precision on standard fonts.

Integrated separation and recognition to reduce errors.

Validated system across 14 fonts and 6 sizes.

Abstract

In this paper, we introduce a model-based omnifont Persian OCR system. The system uses a set of 8 primitive elements as structural features for recognition. First, the scanned document is preprocessed. After normalizing the preprocessed image, text rows and sub-words are separated and then thinned. After recognition of dots in sub-words, strokes are extracted and primitive elements of each sub-word are recognized using the strokes. Finally, the primitives are compared with a predefined set of character identification vectors in order to identify sub-word characters. The separation and recognition steps of the system are concurrent, eliminating unavoidable errors of independent separation of letters. The system has been tested on documents with 14 standard Persian fonts in 6 sizes. The achieved precision is 97.06%.