Source Codes for Multiple Description Coding with Random Quantization

This page describes the source codes for the Multiple Description Coding with Random Quantization (MDRQ) developed in the following papers:

L. Meng, J. Liang, U. Samarawickrama, Y. Zhao, H. Bai and A. Kaup, ‘Multiple Description Coding with Randomly and Uniformly Offset Quantizers’, IEEE Trans. Image Processing, Vol. 23, No. 2, pp. 582-595, Feb. 2014.  PDF

Please click here to download the source codes.

The package includes the following folders:

1.         The package of ‘MDROQ_MDUOQ_filterdesign’ is used to get the optimized filters.


'MDROQ_filter' is used to get the filters of MDROQ,

For 2des and 4des, the command ‘MDROQ_filter(2, 1.5, 0.0679,8, 1, 0)’ can be used to get the optimized filters, which is saved in MDROQ_filter2D_R1.5_p0.0679.dat.

For 3des and 9des, the command ’MDROQ_filter(3, 1.5, 0.1408,8, 1, 0)’ can be used to get the optimized filters, which is saved in MDROQ_filter3D_R1.5_p0.1408.dat.

 

'MDUOQ_filter' is used to get the filters of MDUOQ.

For 2des and 4des, the command ‘MDUOQ_filter(2, 1.5, 0.0679,8, 1, 0)’ can be used to get the optimized filters, which is saved in MDUOQ_filter2D_R1.5_p0.0679.dat.

For 3des and 9des, the command ‘MDUOQ_filter(3, 1.5, 0.161,8, 1, 0)’ can be used to get the optimized filters, which is saved in MDUOQ_filter3D_R1.5_p0.161.dat.

 

Note:  ‘S1k_TRPCSQ_vs_MDRQ.m’ is used to get the Fig.2(c).

 

2.         The package of ‘MDROQ_GM1’is used to get the theoretical and simulation Results for 1-D Data for MDROQ.

(1)    In MDROQ_GM1_2des, ‘MDRQ_correlated.m’ is used to derive the simulation distortion of 2-des MDROQ. ‘MDRQDist.m’ is used to get the theoretical distortion of 2-des MDROQ. ‘show_MDROQ_2des.m’ is used to get the curve in Fig.6(a).

(2)    In MDROQ_GM1_3des, ‘MDRQ_correlated.m’ is used to derive the simulation distortion of 3-des MDROQ. ‘MDRQDist.m’ is used to get the theoretical distortion of 3-des MDROQ. ‘show_MDROQ_3des.m’ is used to get the curve in Fig.6(b).

(3)    In MDROQ_GM1_4des, ‘MDRQ_correlated.m’ is used to derive the simulation distortion of 4-des MDROQ. ‘MDRQDist.m’ is used to get the theoretical distortion of 4-des MDROQ. ‘show_MDROQ_4des.m’ is used to get the curve in Fig.6(c).

 

3.         The package of ‘MDROQ’ is the source codes of MDROQ, which include 2-des MDROQ, 3-des MDROQ, 4-des MDROQ and 9-des MDROQ.

(1)    In MDROQ_2des, the main function is MDROQ_2DES.m. The command ‘MDROQ_2DES('lena.pgm','MDROQ_filter2D_R1.5_p0.0679.dat',40,40.57)’ is used to get the first point in Fig. 7(a) of the paper. The total bit rate of the two descriptions is 0.5bpp.

The file ‘show_lena_2des.m’, ‘show_boat_2des_1bpp.m’ and ‘show_baboon_2des_1bpp.m’ can plot the PSNR curves in Fig. 7. The parameters of q0 and q1 are also included in the file.

(2)    In MDROQ_3des, the main function is MDROQ_3DES.m. The command ‘MDROQ_3DES('boat.pgm', 'MDROQ_filter3D_R1.5_p0.1408.dat', 8,30.41,53.3, 0.1)’ is used to get the first point of MDROQ in Fig. 8(b) of the paper. The total bit rate of the three descriptions is 1.0bpp.

(3)    In MDROQ_4des, the main function is MDROQ_4DES.m. The command ‘MDROQ_4DES('peppers.pgm', 'MDROQ_filter2D_R1.5_p0.0679.dat', 4, 8, 22.2,59.8,1,1)’ is used to get the first point of MDROQ in Fig. 9(c) of the paper. The total bit rate of the four descriptions is 1.0bpp.

(4)    In MDROQ_9des, the main function is MDROQ_9DES.m. The command ‘MDROQ_9DES('couple.pgm', 'MDROQ_filter3D_R1.5_p0.1408.dat',8,12.6,140.3,1 )’ is used to get the first point of MDROQ in Fig. 10(c) of the paper. The total bit rate of the nine descriptions is 2.0bpp.

 

4.         The package of ‘MDUOQ’ is the source codes of MDUOQ, which include 3-des MDUOQ, 4-des MDUOQ and 9-des MDUOQ.

(1)    In MDUOQ_3DES, the main function is MDUOQ_3DES.m. The command ‘MDUOQ_3DES('boat.pgm', 'MDUOQ_filter3D_R1.5_p0.161.dat', 8,31,46, 0.6, 0.1,0.2)’ is used to get the first point of MDUOQ in Fig. 8(b) of the paper. The total bit rate of the three descriptions is 1.0bpp.

The file ‘show_lena_3des.m’, ‘show_boat_3des.m’ and ‘show_goldhill_3des.m’ can plot the PSNR curves in Fig. 8. The parameters of q0 and q1 are also included in the file. The file ‘show_boat_3des.m’ can also plot the expected PSNR curve in Fig. 11(a).

(2)    In MDUOQ_4DES, the main function is MDUOQ_4DES.m. The command ‘MDUOQ_4DES('peppers.pgm', 'MDUOQ_filter2D_R1.5_p0.0679.dat', 4, 8, 23.66,45.04,0.6,1,1)’ is used to get the first point of MDUOQ in Fig. 9(c) of the paper. The total bit rate of the four descriptions is 1.0bpp.

The file ‘show_lena_4des.m’, ‘show_boat_4des.m’ and ‘show_peppers_4des.m’ can plot the PSNR curves in Fig. 9. The parameters of q0 and q1 are also included in the file. The file ‘show_boat_4des.m’ can also plot the expected PSNR curve in Fig. 11(b).

(3)    In MDUOQ_9DES, the main function is MDUOQ_9DES.m. The command ‘MDUOQ_9DES('couple.pgm', 'MDUOQ_filter3D_R1.5_p0.161.dat', 9,8,12.8, 105.75,0.6, 1)’ is used to get the first point of MDUOQ in Fig. 10(c) of the paper. The total bit rate of the nine descriptions is 2.0bpp.

The file ‘show_lena_9des.m’, ‘show_boat_9des.m’ and ‘show_couple_9des.m’ can plot the PSNR curves in Fig. 10. The parameters of q0 and q1 are also included in the file. The file ‘show_boat_9des.m’ can also plot the expected PSNR curve in Fig. 11(c).



shopify analytics