Presently digital document images are so widely used that
innovative techniques are required to organize these document
image databases according to their contents. A scanned document
may consist of graphics such as computer generated
artwork, rendered images of a graphics system, digital images
captured by a camera, logos, stamps, etc. Suitable techniques
to extract graphics and text are required to interpret the
content of a document. Such techniques are also helpful for
an optical character recognition (OCR) system for enhancing
its proficiency. Typically separation of logos and stamps in
document images (Figs. l(a) and (b)) has drawn attention of
several researchers.