AGU - DCT Based High Quality Image Compression
AGU is a high quality DCT 32x32 based lossy image coder.
AGU is FREE for scientific and noncommercial use.
N.N. Ponomarenko, V.V. Lukin, K.Egiazarian, J. Astola, "DCT Based High Quality Image Compression", Proceedings of 14th Scandinavian Conference on Image Analysis, Joensuu, Finland, pp. 1177-1185, June, 2005
The following paper describes a more sofisticated version of AGU with partition schemes. This modification provide results up to 0.6 dB better than for AGU.
N.N. Ponomarenko, K.O. Egiazarian, V.V. Lukin, J.T. Astola, High-Quality DCT-Based Image Compression Using Partition Schemes, IEEE Signal Processing Letters , Vol. 14, Issue 2, February, 2007, pp. 105-108.
A research version of the coder (only for 512x512 grayscale RAW image coding) and some test images are given below:
agu.zip (40 Kb) - AGU.
testset.zip (1.1 Mb) - set of test images (Lenna, Barbara, Baboon, Goldhill, Patterns, Peppers).
PSNR, dB for different bitrates for methods used in the performance analysis:
|
Image |
JPEG
bpp=1 |
JPEG 2000
bpp=1 |
SPIHT
bpp=1 |
X-W 1999
bpp=1 |
GLBT 16x32
bpp=1 |
AGU 32x32
bpp=1 |
AGU MHV
bpp=1 |
JPEG
bpp=0.5 |
JPEG 2000
bpp=0.5 |
SPIHT
bpp=0.5 |
X-W 1999
bpp=0.5 |
GLBT 16x32
bpp=0.5 |
AGU 32x32
bpp=0.5 |
AGU MHV
bpp=0.5 |
JPEG
bpp=0.25 |
JPEG 2000
bpp=0.25 |
SPIHT
bpp=0.25 |
X-W 1999
bpp=0.25 |
GLBT 16x32
bpp=0.25 |
AGU 32x32
bpp=0.25 |
AGU MHV
bpp=0.25 |
|
Lena |
39.15 |
40.33 |
40.46 |
41.21 |
40.43 |
40.50 |
40.85 |
35.88 |
37.27 |
37.25 |
37.87 |
37.33 |
37.51 |
37.86 |
32.49 |
34.15 |
34.14 |
34.76 |
34.27 |
34.50 |
34.75 |
|
Barbara |
36.55 |
38.07 |
37.45 |
39.12 |
38.43 |
39.26 |
39.91 |
31.73 |
32.87 |
32.10 |
34.48 |
33.94 |
34.65 |
35.28 |
27.77 |
28.89 |
28.13 |
30.60 |
30.18 |
30.77 |
31.21 |
|
Baboon |
28.14 |
29.11 |
29.17 |
- |
- |
29.70 |
30.27 |
24.76 |
25.57 |
25.64 |
- |
- |
26.12 |
26.39 |
22.47 |
23.18 |
23.26 |
- |
- |
23.69 |
23.77 |
|
Goldhill |
35.63 |
36.54 |
36.55 |
37.34 |
36.78 |
37.03 |
37.38 |
32.38 |
33.24 |
33.13 |
33.75 |
33.42 |
33.65 |
33.81 |
29.78 |
30.53 |
30.56 |
30.98 |
30.84 |
31.09 |
31.22 |
|
Peppers |
37.33 |
38.17 |
38.37 |
- |
- |
38.33 |
38.91 |
34.82 |
35.80 |
35.82 |
- |
- |
35.55 |
36.04 |
32.15 |
33.54 |
33.51 |
- |
- |
33.32 |
33.95 |
|
Patterns |
30.68 |
35.79 |
28.83 |
- |
- |
38.55 |
43.81 |
22.35 |
28.46 |
20.51 |
- |
- |
30.66 |
32.51 |
17.36 |
22.60 |
16.25 |
- |
- |
25.13 |
26.39 |
JPEG - JPEG with deblocking
JPEG2000 - D. Taubman, M. Marcellin, JPEG 2000: Image Compression Fundamentals, Standards and Practice. Boston: Kluwer, 2002
SPIHT - A. Said, W.A. Pearlman, A new fast and efficient image codec based on the partitioning in hierarchical trees, IEEE Transactions on Circuits Syst. Video Technol.,2 Vol. 6, 1996. - pp. 243-250
X-W 1999 - Z. Xiong, X. Wu, Wavelet image coding using trellis coded space-frequency quantization, IEEE Signal Processing Letters, Vol. 6, Issue 7, July 1999, pp. 158-161
GLBT 16x32 - T.D. Tran, T.Q. Nguyen, A lapped transform progressive image coder, IEEE Proceedings of the International Symposium on Circuits and Systems ISCAS '98, Vol. 4, 1998, pp. 1-4
AGU e <input file> <output file> <Quantization Step> - encode.
AGU d <input file> <output file> - decode.
AGU w <input file> <output file> - decode without deblocking.
AGU p <first RAW file> <second RAW file> - calculation of PSNR.
Smaller value of quantization step provides better quality of decoded image.
Back to main page...
Last changed: 2008-05-02.