Fundamentals

1. Fundamentals

2. Motivation • Issue • Text and lines are indispensable to print quality • Toner overdevelopment causes toner scatter • Toner scatter makes printed pages appear blurred near the edges of text or thin lines • Current solutions • Limit the color gamut of printers • Undesired color shifts and introduction of halftone texture • Compensate for toner overdevelopment by the coring algorithm • Little research due to new exclusive method

3. Toner Scatter EP process Effect of toner scatter

4. Overview of Coring • Technique of decreasing amount of colorant in the interior of text or characters to diminish effect of toner scatter • Optimal coring value: minimum decrease of colorant amount by percentage ∆ of nominal colorant level No coring 18% coring 50% coring Profile of cored line Annulus with10-pixel width Printed at 600 dpi, scanned at 1200 dpi

5. Overview of Coring (cont.) • By utilizing both high and low resolution image data Coring algorithm

6. Goals • Reduce toner scatter in region of interest, while yielding the appearance of sharper edge transition • Overcome disadvantages caused by limiting the color gamut of printers No coring Optimally cored color clipped Printed at 600 dpi, scanned at 1200 dpi

7. Outline • Remote Print Defect Diagnosis • Print Quality Assessment • Fundamentals • Determination on Optimal Coring Values • Preference Experiment • Conclusions • Future Work

8. Determination onOptimal Coring Values

9. Subjective Assessment • Objectives • To evaluate threshold of coring amount by psychophysical experiments • To determine limits of the coring algorithm that may cause unacceptable results • Environment • Equipments • HP Color LaserJet 2605 at printing resolution of 600 dpi • HP Color Laser Glossy Photo Paper 220 g/m2 • Normal ambient lighting condition in controlled laboratory

10. Experimental Design • Pilot experiments • Test stimulus in the shape of annulus • Line widths between 5/600 inch and 50/600 inch • Nominal colorant levels between 90% and 100% in steps of 2.5% • Only combinations of cyan (C) and magenta (M) with same amounts 26 coring levels(Increasing order) 13 coring levels (Increasing order) 7 coring levels(Random order) 7 coring levels (Increasing order) 7 coring levels (Increasing order) 7 colorant levels (100% ~ 70%) 5 colorant levels (100% ~ 80%) 5 Colorant levels (100% to 90%) 5 colorant levels (100% ~ 90%) 5 colorant levels (100% ~ 90%) Pilot test 1 Pilot test 2 Pilot test 3 Pilot test 4 Pilot test 5

11. Experimental Design (cont.) • Test pattern • Fixed nominal colorant level and fixed line width • Five repetitive trials • Randomly distributed over sets of test pages More coring for C and M • Symmetric coring More coring for C • Asymmetric coring More coring for M

12. Experimental Design (cont.) Coring increases Coring of M increases Test pattern Varying nominal colorant level (fixed line width) Coring of C increases Skew mark Color planeregistration bars Calibration patches • Symmetric coring • Decreasing same amountsof C and M colorants • Perception of toner scatter • Asymmetric coring • Decreasing different amountsof C and M colorants • Perception of toner scatter • Color matching

13. Experimental Design (cont.) • Charge to subject • View at normal viewing distance of 10 to 15 inches • Use apparatus to protect test page and facilitate examination of one test pattern at a time • Choose one test stimulus • Symmetric  leftmost one with no toner scatter • Asymmetric  one with no toner scatter and mostclosely matched color of thick border of solid colorant • Disregarded factors: Apparatus for symmetric coring Background noise Jaggedness Color plane misregistration

14. Responses of Subjects Symmetric coring (25 subjects) Asymmetric coring (21 subjects)

15. Optimal Coring Values • More coring for high colorant levels and thin lines • Less coring for low nominal colorant levels and wide lines • More coring for colorant M than colorant C except nominal colorant level 100% with thin line width

16. Comparison with Bernal et al.’s Model • Difference • Line width: very thin and wide line widths (5 and 50 pixels) • Nominal colorant level: low colorant level (90%) • Basis of data for Bernal’s model • Monochrome softcopy experiments using blur in edges and lines • Obtained from two subjects

17. Influence of Edge Directionon Toner Scatter • More coring for vertical than horizontal • Coring amounts for horizontal don’t depend on line width 10 subjects • Duration of toner stealing for vertical is longer than one for horizontal

18. Influence of Edge Directionon Toner Scatter (cont.) Line width: 10 pixels, Nominal colorant level: 100% Printed at 600dpi, scanned at 1200dpi

19. Outline • Remote Print Defect Diagnosis • Print Quality Assessment • Fundamentals • Determination on Optimal Coring Values • Preference Experiment • Conclusions • Future Work

20. Preference Experiment • Objective • To evaluate whether human viewers prefer optimally cored prints over prints without coring in real-world images • Test material • Seven different page contents • Seven different nominal colorant levels (94%~100% in steps of 1%) • Asymmetric coring • Comparison • Control group: 10 subjects who participated in previous experiments • Experimental group: 10 new subjects who don’t know about coring • Charge to subject • Choose a page they prefer over another in terms of color and blur

21. Preference Experiment (cont.) Test set

22. Preference Experiment (cont.) • Results • Overall rate of preference to optimal coring over no coring: 98.57% • Group • Control group: 99.39% • Experimental group: 97.76% • Colorant level

23. Conclusions • Subjects preferred • Less coring as colorant level decreases or line width increases • More coring for colorant M than colorant C • Vertical edges of text or characters provoke toner scatter easily than horizontal edges do • Confirmed that optimally cored prints were overwhelmingly preferred over non-cored prints