The effects of image enhancement in OCR systems: a prototype

The accuracy of the optical character recognition (OCR) systems is highly dependent upon the quality of the image. We investigate and propose solutions to several issues that can arise in the processing of binary images of scanned, typeset text. The issues of concern are image residues from adjacent lines, character touching, boldface character recognition, and text repairing.

[1]  Anil K. Jain,et al.  Address block location on complex mail pieces , 1997, Proceedings of the Fourth International Conference on Document Analysis and Recognition.

[2]  Rangachar Kasturi,et al.  A Robust Algorithm for Text String Separation from Mixed Text/Graphics Images , 1988, IEEE Trans. Pattern Anal. Mach. Intell..

[3]  Ray R. Hashemi,et al.  Hybrid Image Analysis Techniques for Improvement of the OCR Systems End Products , 1998 .

[4]  Daniel P. Lopresti,et al.  Extracting text from WWW images , 1997, Proceedings of the Fourth International Conference on Document Analysis and Recognition.

[5]  Rafael C. González,et al.  Local Determination of a Moving Contrast Edge , 1985, IEEE Transactions on Pattern Analysis and Machine Intelligence.