Browse Prior Art Database

Fast Error Diffusion Method

IP.com Disclosure Number: IPCOM000102180D
Original Publication Date: 1990-Nov-01
Included in the Prior Art Database: 2005-Mar-17
Document File: 2 page(s) / 59K

Publishing Venue

IBM

Related People

Hashihara, H: AUTHOR

Abstract

A technique is disclosed that accelerates the error diffusion method, which is one of the image halftoning techniques without multiplication or division. This technique can be applied for multi-value image, especially for full color image.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 75% of the total text.

Fast Error Diffusion Method

       A technique is disclosed that accelerates the error
diffusion method, which is one of the image halftoning techniques
without multiplication or division.  This technique can be applied
for multi-value image, especially for full color image.

      Fig. 1 shows a configuration of this method.  It consists of
the error buffer for 2 lines, the limiter, the quantization table,
and the diffusion table.

      The error buffer stores cumulative error (S e), and it is
prepared for the current line and next line.  Before processing, both
line buffers are cleared, and after each line processing, the current
line buffer is cleared, and the current line and next line buffers
are swapped.

      The Input data:  x has range (xmin <= x <= xmax). However, the
addition of the cumulative error (x + S e) exceeds this range, and
the entries into the quantization table are increased.  The limiter
is used for reducing this increase.

      The quantization table is used for quantizing data from the
limiter (x') and getting quantized error.  The quantization table has
entries from xmin to xmax, and stores quantized values (q) and
quantized error indexes (e'), determined by quantized error:  e (=x'
- q).

      For full color image processing, bit widths of red, green, and
blue data are different (e.g., RRGGGGBB).  So, exclusive diffusion
tables, which store shifted quantized values, are prepared for each
color, and then final o...