DCT

Visualizing 2D DCT
1D DCT basis set for can be visulized as
For 2D space , we need 64 bases. These bases can be

created using outerproduct of 1D bases.
The bases can be thought of arranged in the form
for convenience
Note that each entry is 8x8 matrix . In total, there are 64 base matrices
For 8x8 2D data we find 2D DCT coefficients by dotproducting the the

data with 64 bases. The result of dotproduct is
another 8x8 matrix of DCT coefficients.
1
Let Then DCT of x is
Note that both x and X are 8x8 matrices
16. For , Create DCT basis set vector (here it is 2D) , DCT_3_4
from 1D basis set for
(basis set matrix created by outer producting 3rd and 4th 1D basis set
vectors in order )
B=idct(eye(8)); b3=B(:,3); b4=B(:,4);

DCT_3_4=b3*b4';
disp(num2str(DCT_3_4,'%.3f '))
0.192 -0.045 -0.227 -0.128 0.128 0.227 0.045 -0.192

0.080 -0.019 -0.094 -0.053 0.053 0.094 0.019 -0.080
2
-0.080 0.019 0.094 0.053 -0.053 -0.094 -0.019 0.080
-0.192 0.045 0.227 0.128 -0.128 -0.227 -0.045 0.192
-0.192 0.045 0.227 0.128 -0.128 -0.227 -0.045 0.192
-0.080 0.019 0.094 0.053 -0.053 -0.094 -0.019 0.080
0.080 -0.019 -0.094 -0.053 0.053 0.094 0.019 -0.080
0.192 -0.045 -0.227 -0.128 0.128 0.227 0.045 -0.192
17. Take M=randi([0 255],8,8) matrix find coefficient corresponding to

DCT_3_4
rng("default")
M=randi([0 255],8,8);
B=idct(eye(8)); b3=B(:,3); b4=B(:,4);
DCT_3_4=b3*b4';
Coef_3_4=dot( M(:), DCT_3_4(:) ); %Vectorise and
take dot product
disp(num2str(Coef_3_4,'%2.3f '))
-57.858
18. Find all 2D DCT coefficients for M=randi([0 255],8,8)
rng("default")
M=randi([0 255],8,8);
D=dct2(M);
disp(num2str(D,'%2.3f ')); % verify DCT_3_4
coefficient
1116.250 102.814 45.745 -99.648 -96.250 -24.298 39.230 -26.612

135.093 58.601 80.169 -21.028 -10.198 109.021 -72.408 -87.192
-17.807 -30.900 99.582 -57.858 3.250 53.300 64.996 -31.462
-21.003 9.791 43.952 22.132 90.329 -105.511 -189.789 -45.653
142.500 2.407 136.573 34.842 1.000 5.638 -16.522 80.689
60.065 -0.230 -112.897 43.329 30.101 -102.171 7.744 4.221
-211.538 -168.923 23.746 21.645 0.007 15.441 98.168 -80.516
36.699 -63.852 -47.301 -135.451 95.246 -15.646 -0.175 -26.563
3
19. Find all 2D DCT coefficients for M=magic(8). what you infer
rng("default")
M=magic(8);
D=dct2(M);
disp(num2str(D,'%2.3f '));
260.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000

0.000 0.000 0.000 0.000 13.990 0.000 0.000 0.000
0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000
0.000 0.000 0.000 0.000 103.888 0.000 0.000 0.000
0.000 1.749 0.000 12.986 0.000 12.783 0.000 0.945
0.000 0.000 0.000 0.000 102.266 0.000 0.000 0.000
0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000
0.000 0.000 0.000 0.000 7.561 0.000 0.000 0.000
Message='DCT allows data compression in transformed

domain';
disp(Message)
DCT allows data compression in transformed domain
20. DCT for image compression (demo without Huffman

coding ).
Demonstrate that Transform in Transform domain, many

coefficients are near zero.
and by cutting of those coefficients do not lead to

percetible change in Reconstructed Image.
rng("default")
M=magic(8);
disp(num2str(M,'%2.0f '))
4
64 2 3 61 60 6 7 57
9 55 54 12 13 51 50 16
17 47 46 20 21 43 42 24
40 26 27 37 36 30 31 33
32 34 35 29 28 38 39 25
41 23 22 44 45 19 18 48
49 15 14 52 53 11 10 56
8 58 59 5 4 62 63 1
D=dct2(M);
threshold = 0.0001;
% Set values to zero if their absolute value is
less than the threshold
D(abs(D) < threshold) = 0;
% Display the modified matrix
disp(num2str(D,'%2.1f '));
260.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

0.0 0.0 0.0 0.0 14.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 103.9 0.0 0.0 0.0
0.0 1.7 0.0 13.0 0.0 12.8 0.0 0.9
0.0 0.0 0.0 0.0 102.3 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 7.6 0.0 0.0 0.0
D_r=idct2(D);
disp(num2str(D_r,'%2.0f '));
64 2 3 61 60 6 7 57
9 55 54 12 13 51 50 16
17 47 46 20 21 43 42 24
40 26 27 37 36 30 31 33
32 34 35 29 28 38 39 25
41 23 22 44 45 19 18 48
49 15 14 52 53 11 10 56
8 58 59 5 4 62 63 1
Projects
Class room Project topics
5
1. Image Compression
2. Watermarking
3. Steganography
4. Audio watermarking
References for doing 2 hour projects.
1. Introduction to Steganography with

MATLAB https://medium.com/@lhagenau/introduction-to-
steganography-with-matlab-d8d2861a3686
2. https://github.com/AhmedAbdElghany97/LSB-Steganography
Matlab code
3. Image Steganography using RSA and Hash
LSB https://in.mathworks.com/matlabcentral/fileexchange/
127893-image-steganography-using-rsa-and-hash-lsb/
4. https://github.com/SaiManojGubbala/Image-Steganography
Good Matlab code
5. Audio_Steganography_in_MATLAB https://github.com/
singhishita/Audio_Steganography_in_MATLAB
6. https://cft.vanderbilt.edu/guides-sub-pages/teaching-in-the-age-
of-ai/

DCT

Uploaded by

Copyright:

Available Formats

DCT

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

DCT

Uploaded by

Copyright:

Available Formats

Visualizing 2D DCT

1D DCT basis set for can be visulized as

For 2D space , we need 64 bases. These bases can be

The bases can be thought of arranged in the form

For 8x8 2D data we find 2D DCT coefficients by dotproducting the the

another 8x8 matrix of DCT coefficients.

Note that both x and X are 8x8 matrices

B=idct(eye(8)); b3=B(:,3); b4=B(:,4);

0.192 -0.045 -0.227 -0.128 0.128 0.227 0.045 -0.192

17. Take M=randi([0 255],8,8) matrix find coefficient corresponding to

18. Find all 2D DCT coefficients for M=randi([0 255],8,8)

1116.250 102.814 45.745 -99.648 -96.250 -24.298 39.230 -26.612

260.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000

Message='DCT allows data compression in transformed

DCT allows data compression in transformed domain

20. DCT for image compression (demo without Huffman

Demonstrate that Transform in Transform domain, many

and by cutting of those coefficients do not lead to

260.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

Class room Project topics

References for doing 2 hour projects.

1. Introduction to Steganography with

You might also like

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.