IDL GUI for Digital Halftoning Final Project for SIMG-726 Computing For Imaging Science Changmeng Liu 2.14.2004.

IDL GUI for Digital Halftoning

Final Project for SIMG-726 Computing For Imaging Science

Changmeng Liu2.14.2004

Outline Objective Digital Halftoning Background Halftoning algorithm Resolution of the display device

and Human Vision MTF and CSF GUI design and demo Conclusion and Future Work

Objective A IDL GUI is implemented to display

the halftone image created by different Digital Halftoning algorithms.

The GUI can simulate the Halftone image in different resolution

Dispersed-dot, Clustered-dot with different orientation, and Random dithering algorithms are implemented

What is Halftone and Why Halftone

A process converts a gray scale image into a bi-level image or more generally to fewer gray levels.

Applications: a. Printing images b. Display images with low-end

display unit c. compressing images & video clips

Continuous Tone image vs. Halftone image

Continuous tone image: every pixel has its continuous valued gray level, for example from 0 to 255

Halftone image: every pixel only has two gray levels: 0 and 1

Example of continuous tone and halftone image

Halftone cell --- Resolution vs. Represent gray level There is a trade off in halftone

technique, the resolution vs. represent gray level

Larger halftone cell can represent more gray level, but larger halftone cell will reduce the resolution of the halftone image.

Note: Error diffusion has no such limitation

Halftone Algorithm The Ordered Dithering Algorithm flow

chart

image

screen

halftone

compare pixel-by-pixel

1 if image > screen0 otherwise

8 bpp

ready for printing on a binary printer8 bpp

1 bpp

Cluster-dot Dithering In its simplest form, a digital version of

analog methods Used in higher resolution printing

devices where system response is nonlinear

Clustering the black pixels yields repeatable, low noise images with acceptable tone reproduction characteristics

Screen Cell The screen is usually constructed

from small rectangular bricks, called cells

98

28

84

14

70 126

56

140

210

196

252

182

238

16822415411242

This screen is constructed from 3x6 cells; the cells are offset by 3 pixels.

Single Cell Construction. The screen on the left with this cell form a 45 degree screen

Halftone Cell Orientation Human Vision have lowest contrast

sensitivity at 45° For monochrome screens (e.g., B&W),

cells are typically nx2n in dimension and offset so the lowest frequency component is at 45°, where the visual MTF is lowest.

m

m

0°

2n

n

45°

Dispersed-Dot Dithering Dispersed-dot produces finer

image details, and its halftone texture pattern contains much higher frequency and therefore is less visible

Bayer’s mask is used as the dispersed-dot dithering mask in this project

Dispersed-Dot Mask Design Recursively Define mask M(k), for

K>0

M(k) has dimensions 2k * 2k

Random Dithering White noise dithering use an

uniformly distributed uncorrelated noise as threshold function.

Random dithering has visible low frequency noise

Resolution of the display device DPI refers to the number of dots (pixels)

per inch on a screen. Macintosh computer (or MacOS

compatible), have a 72 dpi screen. Windows PC have a 96 dpi screen.

Because of this, objects that are displayed under Windows will appear to be 133% of their printed size (at default 72 dpi).

Printer have 72 dpi (default), 300dpi, 600dpi or even higher resolution.

Question Can I Simulate the High Resolution

Halftone Image (300 dpi) with a Low Resolution CRT(96 dpi)?

- No, you can’t display it by just down sampling the high resolution halftone image directly because of the Aliasing.

- But, you can simulate the appearance of the high resolution ht image in CRT.

Human Visual MTF Human

Visual MTF is a kind of low pass filter

Human Visual CSF Contrast Sensitivity

Function of Human Visual system is a band pass filter

From the MTF and CSF, the high spatial frequency in high resolution printer can’t be “seen” by human eye

Simulate different resolution halftone image on CRT Up-sample (or scale) the image to

high resolution size. Applying halftone algorithm Convolve the halftone image with

the low pass filter to simulate what our eye sees

Down-sample the low-passed image to real size

GUI design, flow chart

Dithering

Image Input

Halftone Algorithm Select

Display Resolution Select

Halftone Image Display

Save the Halftone Image

Input Image Display

Output 1, 0° Cluster-dot at 96 dpi

Output 2, 0° Cluster-dot at 300 dpi

Output 3, 45° Cluster-dot at 96 dpi

Output 4, 45° Cluster-dot at 300 dpi

Output 5, Dispersed-dot at 96 dpi

Output 6, Dispersed-dot at 300 dpi

Output 7, Random dithering at 96 dpi

Output 8, Random dithering at 300 dpi

Conclusion and Future Work This GUI can simulate the different

resolution halftone image display Cluster-dot (45° and 0°), disperse-dot, and

random dot is implemented. The convolution kernel can be changed

to fit the human visual MTF and CSF better.

More halftone algorithms can be implemented.

Question?