Center for Signal and Image Processing Seminar

Design and Quality Assessment of Forward and Inverse Error Diffusion Halftoning Algorithms

Prof. Brian L. Evans
Laboratory for Image and Video Engineering
Dept. of Electrical and Computer Engineering
The University of Texas at Austin, Austin, Texas

Collaboration with Dr. Thomas Kite and Mr. Niranjan Damera-Venkata

Thursday, August 27, 3:00 PM
Georgia Center for Advanced Telecommunications Technology Building
Georgia Institute of Technology

Slides in PDF Format


Digital halftoning quantizes a grayscale image to one bit per pixel for printing or display on binary devices. Error diffusion is a halftoning method which employs feedback to preserve the local image intensity and reduce low frequency quantization noise. Error diffusion essentially sharpens the original and adds highpass quantization noise. As a consequence, PSNR is not an appropriate quality measure. As printer resolution increases, images must be enlarged and halftoned, so fast low-cost interpolated halftoning systems become important. However, halftones cannot be scaled, sharpened, rotated, or compressed without causing severe degradation. Degradation can be significantly reduced if the halftone is inverse halftoned (converted to grayscale) before processing. Inverse halftoning essentially blurs the halftone and adds noise. Rehalftoning, which is inverse halftoning followed by halftoning, is a technique under consideration for lossy coding of binary images for the emerging JBIG2 standard.

The talk presents objective measures for the frequency distortion and noise component in forward and inverse halftoning. The measures are based on modeling the quantizer in error diffusion as a gain plus additive noise and modeling the human visual system as a lowpass linear system. We compare competing forward and inverse halftoning schemes in terms of computational complexity, PSNR, and quality. This talk also presents a new algorithm for inverse halftoning, which is based on anisotropic diffusion and is well-suited for implementation in VLSI circuits and embedded software. The new algorithm provides comparable results to the best inverse halftoning method based on wavelet denoising, but at a fraction of the implementation cost. Finally, this talk applies the quantizer model to the improving the design of interpolated halftoning and rehalftoning systems.


Brian L. Evans is an Assistant Professor in the Department of Electrical and Computer Engineering at The University of Texas at Austin, and the Director of the Embedded Signal Processing Laboratory within the Center for Vision and Image Sciences. His research interests include real-time embedded systems; signal, image and video processing systems; system-level design; symbolic computation; and filter design. Dr. Evans has published over 50 refereed conference and journal papers in these fields. He developed and currently teaches EE381K Multidimensional Digital Signal Processing, EE382C Embedded Software Systems, and EE379K Real-Time Digital Signal Processing Laboratory. His B.S.E.E.C.S. (1987) degree is from the Rose-Hulman Institute of Technology, and his M.S.E.E. (1988) and Ph.D.E.E. (1993) degrees are from the Georgia Institute of Technology. From 1993 to 1996, he was a post-doctoral researcher at the University of California at Berkeley with the Ptolemy Project. Ptolemy is a research project and software environment focused on design methodology for signal processing, communications, and controls systems. In addition to Ptolemy, he has played a key role in the development and release of six other computer-aided design frameworks, including the Signals and Systems Pack for Mathematica, which has been on the market since the Fall of 1995. He is an Associate Editor of the IEEE Transactions on Image Processing, a Senior Member of the IEEE, and the recipient of a 1997 National Science Foundation CAREER Award.

Mail comments about this page to