Optical character recognition without segmentation

Abstract

In this study, an off-line optical character recognition scheme is presented. The proposed method performs the recognition by extracting the characters from the whole word, thus, avoiding segmentation process which is the most significant source of error in the recognition process. A set of control points which is defined by the position and attribute vectors are selected as features. In the training mode, each sample character is mapped to a set of control points and is stored in an archive which belongs to an alphabet. In the recognition mode, first, the control points of the input image are extracted. Then, each control point is matched to the control points in the alphabet according to its attributes. During the matching process a probability matrix (PM) is constructed which holds some matching measures (probabilities) for identifying the characters. Experimental results indicate that the proposed method is very robust in extracting the characters from a cursive script and it is insensitive to noise.