Using Orientation Voting to Extract Text Lines with Various Mixed Directions from a Document Image

Abstract

Text line extraction from a document image is a very important task for optical character recognition, document analysis etc. In this paper, a novel approach is presented to extract text lines from a printed or handwritten document image. The document image is binarized at first, and then connected components are detected and consequently character components are collected. For clustering character components into conceptual text lines, a minimum spanning tree (MST) is built based on graph theory. An orientation voting strategy is proposed to compute conceptual consistency of links. After cutting the links with less vote, an initial clustering of character components is obtained. Polynomials are used to model straight or curved text lines, and then polynomials on the same lines are merged to represent a single conceptual text line. Finally, a post-process is applied to delete non-textual components. The experimental results demonstrate that the proposed algorithm performs plausible.