Application of the Factorization Approach in Computer Entry of Printed Text

Abstract

The topic of computer recognition and entry of alpha-numeric information has been known for approximately 30 years. Today, it is still vital, because the problems related to it have not found satisfactory solution. For example, the existing automatic reading devices require special fonts, like OCR -A, OCR -B, CMC -7, E13B, etc.The complete solution of the problem demands a reading system, capable of inputting letters, signs, numerals of arbitrary fonts and height from manuscripts and books. This problem is very complex and very hard to solve. A step towards finding a more general solution is the proposed factorization method for preliminary processing of printed text to be entered in a computer. It makes possible the recognition of letters and signs of different fonts and sizes.Basically, automatic reading devices include an optoelectronic transducer (scanner) and a computer for processing the electric signals. It is known that the higher the resolution of the scanner, the more reliable is the recognition. On the other hand, the necessary capacity of the memory increases with the second power of the resolution. However, the use of special algorithms and preliminary information processing can facilitate to a great extent the process of recognition. The algorithm of the factorization method helps compress and extract the essential initial information for computer recognition, thus reducing the capacity of the needed memory.