Rough-Fuzzy Clustering and M-Band Wavelet Packet for Text-Graphics Segmentation Pradipta Maji, Shaswati Roy, and Malay K. Kundu Machine Intelligence Unit, Indian Statistical Institute, Kolkata, India {pmaji,shaswatiroy t,malay}@isical.ac.in Abstract. This paper presents a segmentation method, integrating ju- diciously the merits of rough-fuzzy computing and multiresolution image analysis technique, for documents having both text and graphics regions. It assumes that the text and non-text regions of a given document are considered to have different textural properties. The M-band wavelet packet is used to extract the scale-space features, which is able to zoom it onto narrow band high frequency components of a signal. A scale-space feature vector is thus derived, taken at different scales for each pixel in an image. Finally, the rough-fuzzy-possibilistic c-means algorithm is used to address the uncertainty problem of document segmentation. The perfor- mance of the proposed technique, along with a comparison with related approaches, is demonstrated on a set of real life document images. 1 Introduction With the advances in information technology, automated processing of docu- ments has become an imperative need. The documents in digitized form require a large amount of storage space, after being compressed using advanced tech- niques. Text-graphics segmentation partitions a document image into distinct regions corresponding to the text and non-text parts facilitating efficient search- ing and storage of the text parts in documents. Many techniques have been proposed to segment the document image into text and non-text regions in the literature [1]. Recently, wavelet techniques have become powerful tools in this domain. Li and Gray [2] have used features based on distribution characteristics of wavelet coefficients in high frequency bands to segment document images into four classes, namely, background, photograph, text, and graph. Kundu and Acharyya [3] proposed a scheme for text-graphics segmentation based on wavelet scale-space features followed by k-means cluster- ing. Lee et al. [4] used an algorithm based on local energy estimation in wavelet packet domain and k-means clustering. In this paper, a text-graphics segmentation method is proposed, which inte- grates the principles of rough-fuzzy computing and multiresolution image anal- ysis technique. This approach is based on the assumption that the text portion of the document image is comprised of one texture class and the non-text part of the other. The M -band wavelet packet (MWP) is used to extract the scale- space features, which offers a richer range of possibilities for document image P. Maji et al. (Eds.): PReMI 2013, LNCS 8251, pp. 530–538, 2013. c  Springer-Verlag Berlin Heidelberg 2013