Computer Graphics Practice

Image processing 1

Computer Graphics Lab | NTUST CSIE

Overview
• Image Processing
• Raster graphics
• 2D Pixel array
• Transform
• Quantized
• Dithering

Computer Graphics Lab | NTUST CSIE 2

Raster Graphics
Pixel image, also known as a raster image, is an image represented using a pixel array.
• Raster, Pixel, Fragment

• Uniform and Regular Graph

Fixed neighbor number, Fixed relative

Simple

Channel

• Format
jpg, png, tga, hdr…

Color depth (8 bit)

Computer Graphics Lab | NTUST CSIE 3

Raster Graphics
It can be viewed as a three-dimensional array.

Width × Height × channel

height

wid
for example, a 100-pixel wide and 80-pixel high image

h t
would have an array size of 100×80×3.

Since each pixel stores a value between 0 and 255, the total
channel
amount of information in the image would be 100×80×3×8
bits.

Computer Graphics Lab | NTUST CSIE 4

Gray Scale
Luminance (B&W TV)

Computer Graphics Lab | NTUST CSIE 5

Quantized
• Quantization, involved in image processing, is a lossy compression technique achieved by compressing a
range of values to a single quantum value. When the number of discrete symbols in a given stream is
reduced, the stream becomes more compressible.

• Uniform

• Populosity

Computer Graphics Lab | NTUST CSIE 6

Uniform Quantization

0 1 … 31 32 … 63 … 255

¿
¿

{
{
0 1 … 7

Computer Graphics Lab | NTUST CSIE 7

Populosity

uniform quantization
256 256 256

32 32 32

Find 256 most popular colors

1……..256 257……32768

Find the closest color by euclidean (L2) distance in RGB space

Computer Graphics Lab | NTUST CSIE 8

Dithering
Dithering is the attempt by a computer program to approximate a color from a mixture of other colors when
the required color is not available.

Computer Graphics Lab | NTUST CSIE 9

Dithering
• Naive Threshold Dithering
• Brightness Preserving Threshold Dithering
• Random Dithering
• Clustered Dithering
• Floyd-Steinberg Dithering
• Color Floyd-Steinberg Dithering

Computer Graphics Lab | NTUST CSIE 10

Naive Threshold Dithering (1/3)
• Deal with hole image by pixel

• Example: threshold is 127, set value to 255 if pixel value bigger than threshold.
Otherwise set value to 0
10

Computer Graphics Lab | NTUST CSIE 11

Naive Threshold Dithering (2/3)
• Deal with hole image by pixel

• Example: threshold is 127, set value to 255 if pixel value bigger than threshold.
Otherwise set value to 0
10 2 5 8 9 0 0
5 128 255 254 255
0 110 210 220 198
7 90 130 168 157
13 75 60 95 43
Src Dst

Computer Graphics Lab | NTUST CSIE 12

Naive Threshold Dithering (3/3)
• Deal with hole image by pixel

• Example: threshold is 127, set value to 255 if pixel value bigger than threshold.
Otherwise set value to 0
10 2 5 8 9 0 0 0 0 0
5 128 255 254 255 0 255 255 255 255
0 110 210 220 198 0 0 255 255 255
7 90 130 168 157 0 0 255 255 255
0 0 0 0 0
13 75 60 95 43
43
Src Dst

Computer Graphics Lab | NTUST CSIE 13

Brightness Preserving Threshold Dithering

Average gray scale

Get threshold

Computer Graphics Lab | NTUST CSIE 14

Brightness Preserving Threshold Dithering

Computer Graphics Lab | NTUST CSIE 15

Brightness Preserving Threshold Dithering
Brightness Preserving Naïve Threshold

Computer Graphics Lab | NTUST CSIE

Random Dithering
• Add a random amount to each pixel before
thresholding
• Typically add uniformly random amount from [-a,a]
Add random value
• Pure addition of noise to the image
[-0.2,0.2]
• For better results, add better quality noise

• For instance, use Gaussian noise (random values sampled

from a normal distribution)

• Should use same procedure as before for

choosing threshold

• Not good for black and white, but OK for more

colors
• Add a small random color to each pixel before finding the
closest color in the table

Computer Graphics Lab | NTUST CSIE 17

Clustered Dithering

if: img[x][y] >= mask[x%4][y%4]

0.7059 0.3529 0.5882 0.2353

0.0588 0.9412 0.8235 0.4118
0.4706 0.7647 0.8824 0.1176
0.1765 0.5294 0.2941 0.6471

Computer Graphics Lab | NTUST CSIE 18

Floyd-Steinberg Dithering
• Start at one corner and work through image pixel by pixel
• Usually scan top to bottom in a zig-zag

• Threshold each pixel

• Compute the error at that pixel: The difference between what should be there and what you did put
there
• If you made the pixel 0, e = original; if you made it 1, e = original-1

• Propagate error to neighbors by adding some proportion of the error to each unprocessed neighbor
• A mask tells you how to distribute the error

• Easiest to work with floating point image

• Convert all pixels to 0-1 floating point

Computer Graphics Lab | NTUST CSIE 19

Floyd-Steinberg Dithering

Computer Graphics Lab | NTUST CSIE

Floyd-Steinberg Dithering

Computer Graphics Lab | NTUST CSIE 21

Color Floyd-Steinberg Dithering
All the same techniques can be applied, with some modification

Example: 8 bit framebuffer

Set color map by dividing 8 bits into 3,3,2 for RGB

Blue is deemphasized because we see it less well

Dither RGB separately

0 36 73 109 146 182 219 255
Works well with Floyd-Steinberg
0 36 73 109 146 182 219 255
Generally looks good
0 85 170 255
Example is Floyd-Steinberg:
Uniform color table

Error is difference from nearest color in the color table 7/16

Error propagation same as that for greyscale
3/16 5/16 1/16
Each color channel treated independently

Computer Graphics Lab | NTUST CSIE 22

Color Floyd-Steinberg Dithering
Origin Color Floyd Uniform Quant

Computer Graphics Lab | NTUST CSIE

Dithering

• Pixel Operations

• Block Operations

• Neighborhood Operations

Computer Graphics Lab | NTUST CSIE 24

Command
• ScriptHandler.cpp

switch (command)
{
case LOAD:
{
if (pImage)
delete pImage;
char* sFilename = strtok(NULL, c_sWhiteSpace);
bResult = (pImage = TargaImage::Load_Image(sFilename)) != NULL;

if (!bResult)
{
if (!sFilename)
cout << "Unable to load image: " << endl;
else
cout << "Unable to load image: " << sFilename << endl;

bParsed = false;
}// if
break;
}// LOAD

Computer Graphics Lab | NTUST CSIE

Application
• TargaImage.h

Computer Graphics Lab | NTUST CSIE 26

Application
• Image

• RGBA

• 8bit each channel

• Unsigned char (0~255)

Computer Graphics Lab | NTUST CSIE 27

Homework
• 本次作業開始，未按照上傳格式以 0 分計算 !

• Grayscale

• Uniform Quantization
Use the uniform quantization algorithm to convert the current image from a 24 bit color image to an 8 bit color image.

Use 4 shades of blue, 8 shades of red, and 8 shades of green in the quantized image.

• Clustered Dithering
Dither an image to black and white using cluster dithering with the matrix shown below. The image pixels should be
compared to a threshold that depends on the dither matrix below. The pixel should be drawn white if: I[x][y] >=mask[x%4][y
%4]. The matrix is 0.7059 0.3529 0.5882 0.2353
0.0588 0.9412 0.8235 0.4118
0.4706 0.7647 0.8824 0.1176
0.1765 0.5294 0.2941 0.6471
Computer Graphics Lab | NTUST CSIE 28