|
Applied Math Seminar
Spring Quarter 2002
3:15 p.m.
Sloan Mathematics Corner
Building 380, Room 380-C
Friday, May 3, 2002
Kathrin Berkner
Ricoh Innovations, Inc.,
Color Image Processing Group
Enhancement of scanned documents in a digital copier
using properties of wavelets in smoothness spaces
Abstract:
|
|
The development of wavelet theory and applications over
the last decade has shown great success especially in the areas of image
compression and denoising. One of many explanations for that success is
that wavelet bases adapt locally in smooth and non-smooth parts due to
the theoretical link between wavelets and smoothness spaces.
Another field of image processing that is very important for industrial
applications is the field of document enhancement and restoration. The
problem of document enhancement arises in many applications such as scanners,
copiers and digital cameras. A desired enhancement step often includes
a denoising and a deblurring step. We use knowledge of the same smoothness
spaces that lead to advantages in wavelet-based image compression and
denoising to model characteristics of documents, the image capturing process
and the enhancement process in order to perform document enhancement for
a copier entirely in the wavelet domain. The individual processing steps
are designed to be of low computational complexity since - due to computational
constrains - the industry is still slow in substituting traditional Fourier
techniques with wavelet-based techniques.
|