Evaluating Systematic Print Mottle

1
Evaluating Systematic Print Mottle

• Carl-Magnus Fahlcrantz
• _at_ the 11th IPGAC,
• Bordeaux, France, October 1-3, 2002

2
Print Mottle

• Inhomogeneities in the Print due to Reflectance
  Variations.

3
Systematic Print Mottle

• Disturbances in the print that are experienced
  ordered in some sense.

• Occurrence Contemporary Interest
• Digital Printing seems to produce more systematic
  noise than traditional printing methods.

4
Human Visual System A Briefing Contrast
Sensitivity Detection of Structure

• Contrast Sensitivity
• Different perceptibility to noise of different
  sizes.

• Detection of Structure
• Evolution made the Human Visual System
  hyper-sensitive to ordered structures in the
  visual field.

5
Perceiving Mottle An Overview

• Implications
• Evaluation instruments that only considers the
  variation in the print, and not the dominating
  size and the order/structure of the noise,
  underestimate the impact of the mottle on Print
  Quality.

6
Instrumental Mottle Evaluation An Overview of
the Main Idea
4.Estimate Image Noise
1.Scanned Image
2.Filter Image to compensate for Contrast
Sensitivity
3.Filter Image to compensate for Order/Structure
7
Component One Amount of Variation

• Spatial or Frequency Domain?
• Psychophysical Evidence Eye - Frequency Analysis
• Convenient way of doing filtering operations.
• Frequency Domain - Natural Choice.
• Use Energy in FFT Power Spectrum to estimate the
  Variation in the print for different
• Wavelengths
• Orientations

FFT Power Spectrum
8
Component Two Contrast Sensitivity

• Filter the Power Spectrum so that it corresponds
  to the variation detected by the eye.
• Mathematical Approximation available (Barten,
  1999, among others).
• ? Geye(u)

9
Component Three Texture Detection

• Estimation of the order/structure Gain
• Calculate the 2ND Moment locally in the spectrum,
  (Jernigan dAstous, 1984 Nguyen Jordan,
  1989, among others)
• Filter the Spectrum so that the energy in areas
  with a large 2ND Moment (or Chi Square Measure)
  is enhanced.

? Gtexture (Weighted Normalized Chi Square
Measure)
10
Integration

• Sum up all the filtered energy in the Spectrum
  that falls within the visual area.
• ? Single estimate of the Perceived Mottle in the
  Print!

11
Testing the Model using two Sets of Samples

• Digitally Simulated Set
• 12 grayscale samples showing various degrees of
  Systematic Mottle.
• Printed under very stringent conditions with a
  high-quality inkjet printer.
• ? Test the theory with a Challenging Evaluation.
• Set of Sample from Conventional Presses
• 40 samples (K40) from 7 different presses,
  including
• Offset
• Flexography
• Digital Liquid Toner
• Dry Toner presses
• ? Test Practical Applicability.

12
Visual and Alternative Evaluation

• Visual Evaluation
• Simulated Samles Print Mottle Level
• Conventional Prints Print Quality
• Comparison with alternative Instrument
• Traditional STFI Print mottle Evaluation by
  band-pass (1-8 mm) image analysis method
  (Johansson, 1993)
• Very good agreement between measured stochastic
  mottle and Visual Print Mottle Evaluation.
• However Do not care for Contrast Sensitivity
  Order/Structure.

13
Results Simulated Test Set
R 0.10
R 0.97
14
Results Conventional Prints
R 0.90
R 0.75
R 0.80
R 0.26
15
Conclusions

• The results from both evaluations suggests that
  the new model outperforms the traditional model.
• ? Potential candidate for the Evaluation of Print
  Mottle.
• The performance difference, as predicted,
  especially clear when mottle is ordered.
• ? Very potential candidate if the prints show
  Systematic Print Mottle.

16
Acknowledgements

• I would like to thank the following organisation
  and persons for their part in this project
• S2P2 Digital Printing Research Program
• Siv Lindberg
• Don Donderi
• Per-Åke Johansson

Carl-Magnus Fahlcrantz, cmf_at_stfi.se