Real-timeDynamicOfflineTextRecognitionsystem.docVIP

Real-time Dynamic Offline Text Recognition System Muhammad Faisal Zafar1 FSKSM UTM, 81310 Skudai, Johor, Malaysia. Dzulkifli Mohamad2 FSKSM UTM, 81310 Skudai, Johor, Malaysia Ikramul Haq3 COMSAT Institute of Technology, Islamabad Pakistan Abstract A real-time dynamic system for text recognition has been developed. There is no need of prior training. In the beginning, the system detects the text and asks about the name of every new character (alphabet, numeral or special characters) only once, then it automatically recognizes the character when it appears again in the text. The space between the characters has also been detected. Counter propagation neural network has been used for recognition purpose. Every character is grabbed by its extreme coordinates from left /right and top/bottom and a grid, with specific rows and columns, is drawn which covers the whole character. The algorithm automatically adjusts the size of grid and its constituents according to the dimensions of the character. Then it searches the presence of character pixels in every box of the grid. The boxes found with character pixels are considered “on” and the rest are marked “off”. A binary string of each character is formed locating the “on” and “off” boxes (named as character digitization) and presented to the neural network input for online training and recognition purposes. The system also calculates the rows in the document and total characters in every row. The main advantage is an optimized online training domain i.e. only those characters are trained online which are present in the text. Keywords: Optical character recognition, text detection, character digitization, counterpropagation neural networks, extreme coordinates. 1. Introduction Optical character recognition (OCR) has been a topic of interest since possibly the late 1940s when Jacob Rabinow started his work in the field [ERIC 1992]. The earliest OCR machines were primitive mechanical devices with fairly high failure rates. A