Fuzzy Runlength in Document Image Processing
Zhixin Shi, Venu Govindaraju · 2004
In this paper we introduce a novel document image processing technique called fuzzy runlength map(FRM). The technique is designed for geometrical layout analysis and document objects such as text lines, words or graphics locations. A FRM of a document image is a transform of the image. It captures the multi-resolution connectivity features for the basic construction elements ranging from text words to text blocks. With a connectivity level setting determined manually or automatically, FRM reveals the geometrical structure of a document at the level by dening the document objects such as text words or text lines into groups. The proposed technique was tested on a variety of complex document images for document skew detection, text line seperation and other graphics object locations. The document images in the test include postal parcel images and historical handwritten documents such as Newton’s and Galileo’s manuscripts. Comparing with the document image processing techniques published in the literature, our proposed method is superb in both eciency and the strong ability to deal with the complexity of document images including handwritten, machine print text and graphics such as barcode, stamps, logos and other graphical gures.