Morphology-Based Text Line Extraction

Abstract

This paper presents a morphology-based text line extraction algorithm for extracting text regions from cluttered images. A novel morphology-based method is firstly proposed to extract important contrast features as filters to find all possible text region candidates. The contrast feature is robust to lighting changes and invariant to different image transformations like scaling, translation, and skewing. A text line selection process is then applied to locating all possible text-analogue segments from the analyzed image. An x-projection technique is used to extract different geometrical properties of characters from the above text-analogue segments. In addition, a verification process is applied to verifying all extracted text candidates according to the regularities of characters embedding in them. Due to noise, a text region is often fragmented to different pieces of segments. For tackling this problem, a novel recovery algorithm is proposed for locating a complete text line. Since the morphology-based method can significantly reduce the number of candidates, all desired text lines can be efficiently and effectively located. The average accuracy of the proposed system is nearly 94%. Experimental results show that the proposed method improves the state-of-the-art work in terms of effectiveness and robustness for text region detection.