Memory efficient error diffusion and halftone texture characterization

Ti-chiun Chang, Purdue University

Abstract

Because of its good image quality and moderate computational requirements, error diffusion has become a popular halftoning solution for desktop printers, especially those that use inkjet technology. In this dissertation, we will develop memory efficient error diffusion algorithms which greatly accelerate halftoning process for very low cost or large format printers, or reduce implementation cost of generic printers. In chapter 1, we first develop a reduced lookup table implementation of tone-dependent error diffusion (TDED). This algorithm relieves the requirement to store the tone-dependent parameters and halftone bitmap in standard TDED. Secondly, we introduce a new serial block-based approach to error diffusion. This depends on a novel intrablock scan path and the use of different parameter sets at different points along that path. We show that serial block-based error diffusion reduces off-chip memory access by a factor equal to the block height. In chapter 2, we consider fixed-rate scalar quantization of the accumulated diffused error (ADE) in error diffusion. We explain the encoding and decoding procedures which simultaneously operate with error diffusion, and show that the required on-chip random access memory (RAM) for storing the ADE can be reduced by a factor of 2, 3, or 4. We demonstrate 5 methods for designing different quantizers. In chapter 3, we develop a new framework based on local sequency analysis and a quasi filter bank structure in order to characterize halftone textures. Our framework provides a simple mean to decompose a halftone image into subband images, based on which we can easily reconstruct the original halftone and formulate texture characterizations.

Degree

Ph.D.

Advisors

Allebach, Purdue University.

Subject Area

Electrical engineering

COinS