Title :
Thresholding of images by Shannon´s criterion
Author_Institution :
Novosibirsk State Tech. Univ., Russia
Abstract :
Binary classification of images with the minimum information loss is considered. A digital algorithm for printed texts and an experimental method estimating of recognizing quality is proposed.. Some examples are given.
Keywords :
image classification; information theory; optical character recognition; probability; OCR; Shannon´s criterion; binary classification; digital algorithm; image thresholding; printed texts; Gaussian distribution; Mutual information; Probability; Quantization; RF signals; Radio frequency; Signal to noise ratio; Uncertainty;
Conference_Titel :
Electronic Instrument Engineering Proceedings, 2002. APEIE 2002. 2002 6th International Conference on Actual Problems of
Print_ISBN :
0-7803-7361-8
DOI :
10.1109/APEIE.2002.1075809
