Browse Prior Art Database

Digital Image Processing Algorithm for High Frequency Information

IP.com Disclosure Number: IPCOM000044557D
Original Publication Date: 1984-Dec-01
Included in the Prior Art Database: 2005-Feb-06
Document File: 2 page(s) / 34K

Publishing Venue

IBM

Related People

Yeskel, FJ: AUTHOR

Abstract

This article describes a processing technique which identifies areas of high frequency information and provides a direct method for making the print or no-print decision within those areas. For each pixel or pel, a quantity called "curvature" is calculated. The words pixel and pels are used interchangeably to mean a "picture element." The curvature information is utilized to distinguish high frequency information in a mixed format document. Consider a one-dimensional vector of pixel video values where "P" is a pixel index: ...,V(P-1), V(P), V(P+1),...

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 57% of the total text.

Page 1 of 2

Digital Image Processing Algorithm for High Frequency Information

This article describes a processing technique which identifies areas of high frequency information and provides a direct method for making the print or no-print decision within those areas. For each pixel or pel, a quantity called "curvature" is calculated. The words pixel and pels are used interchangeably to mean a "picture element." The curvature information is utilized to distinguish high frequency information in a mixed format document. Consider a one-dimensional vector of pixel video values where "P" is a pixel index: ...,V(P-1), V(P), V(P+1),... The one-dimensional curvature of pel P, computed at the radius R, is defined as: C(P,R) = V(P) - (!V(P-R) + V(P+R)1/2) Thus, the one-dimensional curvature can be seen to be a measure of difference between a linear or average midpoint between the video values of pels (P-R), (P+R), and the actual midpoint video value, V(P). The reason for the name "curvature" is apparent. It is a measure of the departure from linearity between the two pels. Note that pels P, (P-R), and (P+R) are adjacent only if R=1. The extension to the more useful and practical case of two dimensions is straightforward. Define the two-dimensional discrete spatial curvature about pel (P,L) at the radius R as follows: C(P,L,R) =

V(P,L) - A(P,L,R) or C = V-A where: P = pel coordinate L = line coordinate R = discrete radius C(P,L,R) =

discrete spatial curvature about location (P,L) at rad...