Date of Award

4-1997

Degree Name

Bachelor of Science

Department

Paper Science and Engineering

First Advisor

Dr. Brian Scheller

Abstract

Correlation between image analysis measurements and visual ranking of ink jet printing was established. Human rating was determined by the use of an all-pair comparison. A camera driven image analysis system and NIH image analysis software with a scanner was used to analyze printed samples. Three sets of printed samples were evaluated for print density. and single dot attributes.

Analyzing the mean gray scale value (GSV) as a tool for evaluating ink jet print quality is limited. Large differences between samples will contribute to the perception of quality, but small differences are overruled by other factors.

Using the standard deviation of the mean GSV to incorporate mottle effects into image analysis is promising. However, a small amount of mottle will not correlate to print quality.

Object area may be a useful tool for evaluating print quality. Sample set 1, where the printer was held constant, showed decent correlation. This relationship should be further explored.

Perimeter is another good tool for interpreting print quality. As a direct measure of feathering and wicking, perceived print quality has a direct relationship of measured print quality. Once again, the set of samples in which the printer was held constant showed good correlation.

Roundness, being a normalized parameter, is a good indicator of print quality if there is a noticeable variation between different samples.

A rating equation was established which gave excellent correlation between measured print quality and perceived print quality. The equation is as follows:

R1 = (1/P = 1/SD) x C1

R2 = (1000/P = 1/SD) x C2

Where: P=perimeter; SD=standard deviation of the GSV; C1, C2=scaling constants

When applied to data gathered during the course of this project, the regression analysis, when compared to the all pair comparison values, generated an r2 value of up to 0.92. When applied to literature values, the regression value was 0.96.

Share

COinS