Radon transformation for Arabic character recognition
Eman Abdulaziz, Khalil Alsaif · 2008
In this paper an algorithm adopted the properties given by Radon transformation studied to be used by an optical character recognition system for printed Arabic characters. The algorithm gets a high difference between the character features which was extracted by Radon transformation, when it was applied on isolated Arabic characters. A preprocess was needed to be done on the character image before applying Radon transformation. The preprocess steps include the following: 1) truncation for all unused area. 2) size normalization. 3) character skelton by applying thinning continuously. And a database file was created to hold the following features: - The printed character - character image - Radon coordinates - number of pixels individual printed characters are tested. Even the tested characters were so closed to each other, a high rate of recognition reached.