Browse Prior Art Database

Continuous To Halftone Image conversion with Controlled Dot Clustering to Minimize Scanned Reproduction Data Loss

IP.com Disclosure Number: IPCOM000088516D
Original Publication Date: 1977-Jun-01
Included in the Prior Art Database: 2005-Mar-04
Document File: 4 page(s) / 70K

Publishing Venue

IBM

Related People

Stucki, P: AUTHOR

Abstract

This article describes a technique for the conversion of a gray scale image formed from heterogeneously sized and clustered dots into a visually perceived equivalent image. This equivalent image (termed "halftone image") is made from dots (black/white) of uniform size. The method of clustering the uniformly sized dots includes the controlled use of black or white dots to fill in the spaces around object boundaries. This permits data compression of the halftone image whose uniformly sized dots are clustered. The method advantages are (a) minimizes the loss of information (resolution) upon making a scanned and thresholded reproduction from the halftone image, (b) limits the information loss to the making of the first copy, and (c) exhibits a high data compression potential.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 47% of the total text.

Page 1 of 4

Continuous To Halftone Image conversion with Controlled Dot Clustering to Minimize Scanned Reproduction Data Loss

This article describes a technique for the conversion of a gray scale image formed from heterogeneously sized and clustered dots into a visually perceived equivalent image. This equivalent image (termed "halftone image") is made from dots (black/white) of uniform size. The method of clustering the uniformly sized dots includes the controlled use of black or white dots to fill in the spaces around object boundaries. This permits data compression of the halftone image whose uniformly sized dots are clustered. The method advantages are (a) minimizes the loss of information (resolution) upon making a scanned and thresholded reproduction from the halftone image, (b) limits the information loss to the making of the first copy, and (c) exhibits a high data compression potential.

There exist basically two approaches to scan a halftone representation of a photograph. These are high resolution scanning and thresholding for binary representation, and high resolution scanning, 2dimensional digital filtering, subsampling at a lower resolution and quantization for continuous-tone representation. The scanning resolution required has to be high enough in order to avoid any appearance of Moire patterns. It is generally assumed that the high resolution sampling rate must be about eight times the halftone screening density. The factor of eight is experimentally determined and may vary according to the quality of the scanning/reproduction equipment used.

In order to obtain a more efficient description of a rescanned halftone representation of a photograph, high resolution scanning, 2dimensional digital filtering, subsampling at a lower resolution and quantization for continuous-tone representation is often proposed. In practice, a Gaussian or raised cosine smoothing function is generally used. In order to avoid the formation of Moire patterns, the distance between the half-amplitude points of the Gaussian or the raised cosine smoothing function has to be greater than or equal to the halftone screening resolution R(s). This fact leads to the conclusion that a Moire-free reproduction of a halftone original using high resolution scanning, 2-dimensional smoothing and subsampling tends to degrade the resolution of a copy substantially. This can be explained as follows: The scanning process can be expressed as a convolution of the image f(x,y) with a smoothing function h(x,y) and the resulting frequency spectrum G'(u,v) = F(u,v).H(u,v). In the case of a 2- dimensional Gaussian-shaped aperture function h(x,y), H(u,v) also has a Gaussian-shaped frequency spectrum distribution. Therefore, no matter what the frequency spectrum of F(u,v) looks like, the distribution of G'(u,v) is getting narrower and narrower after each copy-cycle and consequently the image quality becomes low-pass degraded accordingly (linear filtering). Generation of Uniform Sized...