I.T.: A Statistically based, Highly Accurate Text-Line Segmentation Method

This paper describes a text-line identification and seg-mentation technique that is probability based, where all probabilities are estimated from an extensive training set of various kind of measurements of distances between the terminal and non-terminal entities and between the text-line and the text-block entities with which the algorithm works. The off-line probabilities estimated in the training then drive all decisions in the on-line segmentation algo-rithm. On the UW-III database of some 1600 scanned document image pages, having some 105,020 text lines, the algorithm identifies and segments 104,773 correctly, an accuracy of 99.76%. 1