Scribd
Upload
224 views4 pages

JPEG Compression for Engineers

The document describes the steps of color JPEG compression on an image. It reads in an image, converts it from RGB to YUV color space, divides the image into 8x8 blocks, applies discrete cosine transform (DCT) and quantization to each block, then performs dequantization, inverse DCT and combines blocks to reconstruct the compressed image. It notes that the DCT and IDCT results are in decimal form and it is unsure how to display the reconstructed image. The document uses Matlab functions to implement the DCT and IDCT transforms.

Uploaded by

Ozora Tuh Shinta
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOC, PDF, TXT or read online on Scribd
224 views4 pages

JPEG Compression for Engineers

The document describes the steps of color JPEG compression on an image. It reads in an image, converts it from RGB to YUV color space, divides the image into 8x8 blocks, applies discrete cosine transform (DCT) and quantization to each block, then performs dequantization, inverse DCT and combines blocks to reconstruct the compressed image. It notes that the DCT and IDCT results are in decimal form and it is unsure how to display the reconstructed image. The document uses Matlab functions to implement the DCT and IDCT transforms.

Uploaded by

Ozora Tuh Shinta
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOC, PDF, TXT or read online on Scribd

% Color JPEG Compression Standart

clear all;clc;close all;


%1. Read Input Image %
im = imread('lena.tif');
[rows, cols, color_space] = size(im);
%2. Convert RGB to YUV (Y'CbCr) %
RGB_image = double(im);
Y = 0.299*RGB_image(:,:,1) + 0.587*RGB_image(:,:,2) + 0.114*RGB_image(:,:,3)
+ 16; % Y = Y' %
U = -0.169*RGB_image(:,:,1) - 0.331*RGB_image(:,:,2) + 0.500*RGB_image(:,:,3)
+ 128; % U = Cb ~ Y-B %
V = 0.500*RGB_image(:,:,1) - 0.419*RGB_image(:,:,2) - 0.081*RGB_image(:,:,3)
+ 128; % V = Cr ~ Y-R %
YUV_image(:,:,1) = Y;
YUV_image(:,:,2) = U;
YUV_image(:,:,3) = V;
%3. Blocking 8x8 %
MM = ceil(cols/8); % MM = Horizontal Block %
NN = ceil(rows/8); % NN = Vertical Block
%
Y_Block = zeros(8,8,MM*NN);
U_Block = Y_Block;
V_Block = Y_Block;
n=0;
for i=0:NN-1
for j=0:MM-1
n=n+1;
Y_Block(:,:,n) = YUV_image(j*8+1:j*8+8, i*8+1:i*8+8, 1);
U_Block(:,:,n) = YUV_image(j*8+1:j*8+8, i*8+1:i*8+8, 2);
V_Block(:,:,n) = YUV_image(j*8+1:j*8+8, i*8+1:i*8+8, 3);
end;
end;
%4. Discrete Cosine Transform (DCT) %
Gy = zeros(8,8,MM*NN);
Gu = Gy;
Gv = Gy;
for i=1:MM*NN
Gy(:,:,i) = dct2(Y_Block(:,:,i)-128);
Gu(:,:,i) = dct2(Y_Block(:,:,i)-128);
Gv(:,:,i) = dct2(Y_Block(:,:,i)-128);
end;
%5. Quantization %
% Luminance Quantization Tables
Ql_Table = [
16 11 10 16 24 40
12 12 14 19 26 58
14 13 16 24 40 57
14 17 22 29 51 87
18 22 37 56 68 109
24 35 55 64 81 104
49 64 78 87 103 121
72 92 95 98 112 100

(8x8) %
51 61
60 55
69 56
80 62
103 77
113 92
120 101
103 99

];
% Chrominance Quantization
Qc_Table = [
17 18 24 47 99
18 21 26 66 99
24 26 56 99 99
47 66 99 99 99
99 99 99 99 99
99 99 99 99 99
99 99 99 99 99
99 99 99 99 99
];

Tables (8x8) %
99
99
99
99
99
99
99
99

99
99
99
99
99
99
99
99

99
99
99
99
99
99
99
99

By = zeros(8,8,MM*NN);
Bu = By;
Bv = By;
for i=1:MM*NN
By(:,:,i) = round(Gy(:,:,i)./Ql_Table);
Bu(:,:,i) = round(Gu(:,:,i)./Qc_Table);
Bv(:,:,i) = round(Gv(:,:,i)./Qc_Table);
end;
% Proses Decoding %
%6. Dequantization

Dy = zeros(8,8,MM*NN);
Du = Dy;
Dv = Dy;
for i=1:MM*NN
Dy(:,:,i) = round(By(:,:,i).*Ql_Table);
Du(:,:,i) = round(Bu(:,:,i).*Qc_Table);
Dv(:,:,i) = round(Bv(:,:,i).*Qc_Table);
end;
%7. Invers Discrete Cosine Transform (IDCT)
Iy =
Iu =
Iv =
for

zeros(8,8,MM*NN);
Iy;
Iy;
i=1:MM*NN
Iy(:,:,i) = idct2(Dy(:,:,i));
Iu(:,:,i) = idct2(Du(:,:,i));
Iv(:,:,i) = idct2(Dv(:,:,i));

end;

%8. Tambahkan dengan nilai skalar 128 %


Sy = zeros(8,8,MM*NN);
Su = Sy;
Sv = Sy;
for i=1:MM*NN
Sy(:,:,i) = (Iy(:,:,i)+128);
Su(:,:,i) = (Iu(:,:,i)+128);
Sv(:,:,i) = (Iv(:,:,i)+128);
end;

Keterangan :

Hasil DCT & IDCT hanya bisa dilihat dalam bentuk desimal. Belum
mengetahui bagaimana cara menampilkan dalam bentuk image (foto).
Untuk rumus DCT & IDCT mengambil langsung dari fungsi Matlab yaitu dct2
& idct2.

You might also like