Method and system for estimating objective quality of compressed video data

• US 6,810,083 B2
• Filed: 11/16/2001
• Issued: 10/26/2004
• Est. Priority Date: 11/16/2001
• Status: Expired due to Term

• Current Assignee: Uniloc 2017 LLC (SoftBank Group Corp.)
• Original Assignee: Koninklijke Philips Electronics N.V. (Koninklijke Philips N.V.)
• Inventors: Caviedes, Jorge, Chen, Yingwei
• Primary Examiner(s): Le, Vu
• Application Number:
  US09/996,003
• Publication Number:
  US 20030112333A1
• Time in Patent Office: 1,075 Days
• Field of Search: 375/240.12, 375/240.13, 375/240.18, 375/240.19, 375/240.24, 375/240.25, 375/240.26, 375/240.27, 375/240.29, 375/243, 375/240.03, 348/420.1, 348/425.1, 348/425.2, 348/408.1, 348/671, 382/235, 382/237, 382/268, 382/276, 382/251, 725/107, 725/130, 725/150
• US Class Current: 375/240.25
• CPC Class Codes: H04N 17/004 for digital television systems
• CPC Class Codes: H04N 19/60 using transform coding
• CPC Class Codes: H04N 19/85 using pre-processing or post-processing specially adapted for video compression

Abstract

The present invention relates to a method and system for evaluating the quality of encoded video data without gaining access to the source data or the compressed video bitstream. The system is configured to decode compressed video data using an MPEG decoder to produce decompressed video data. The decoded data is analyzed to determine whether the decompressed video data is intra-coded. If so, a discrete cosine transform (DCT) is performed to produce a set of DCT coefficients for at least one AC frequency band in the decompressed video data. At the same time, quantization matrix data of a frame of the decompressed video data as well as a quantizer scale for each block of the decompressed video data are extracted. Thereafter, the variance of the converted DCT coefficients is obtained, and then an average quantization error for each set of said DCT coefficients is determined based on the variance, the quantization matrix, and the quantizer scale. Lastly, a peak signal to noise ratio (PSNR) is calculated based on the resultant average quantization error.

Claims

1. A method for evaluating the quality of encoded video data, the method comprising the steps of:

decoding at least a substantial portion of said encoded video data to produce decompressed video data including a plurality of blocks;

performing a discrete cosine transform (DCT) on said decompressed video data to produce a set of DCT coefficients for at least one AC frequency band;

simultaneously, extracting quantization matrix data for the at least one AC frequency band and extracting a quantizer scale for each block of said decompressed video data;

estimating a variance of said DCT coefficients;

determining an average quantization error for each set of said DCT coefficients based on said variance, said quantization matrix, and said quantizer scale; and

,

calculating a peak signal to noise ratio (PSNR) based on said average quantization error.

2. The method of claim 1, wherein said quantizer step size is calculated by averaging the AC coefficients in each of said decompressed video data as follows:

${\stackrel{\_}{C}}_{i,j}=\frac{C_{i,j}\times <br><br>16}{W_{i,j}},$

where C_{i, j} represents the (i, j)th AC coefficient in the current block, {overscore (C)}_{i, j} represents the normalized AC coefficient, and W_{i, j} represents the (i, j)th quantization matrix.

3. The method of claim 2, wherein the variance of the coefficients (λ

²_ij) is determined by the following equation;

${\mathrm{\lambda <br><br>}}_{i,j}^2=\frac{\underset{b=1}{\overset{N}{\sum <br><br>}}<br><br>{\left(C_{i,j}^b\right)}^2}{2<br><br>N},$

where C^b_ij represents the (i, j)th AC coefficient in each block (b), and N represents the total number of blocks.

4. The method of claim 3, wherein the average quantization error (D) is computed as follows:

$D=\frac{\underset{i=0,j=0}{\overset{I=7,j=7}{\sum <br><br>}}<br><br>D_{i,j}}{64},$

and D_{i, j}=2λ

²_{i, j},

where λ

²_ij represents the variance of the coefficients (λ

²_ij), and D_{i, j} represents the quantization error for the (i, j)th AC coefficient in each block.

5. The method of claim 4, wherein said PSNR is calculated as follows:

$\mathrm{PSNR}=10<br><br>\text{\hspace{1em}<br><br>}<br><br>{\log }_{10}<br><br>\frac{255\times <br><br>255}{D},$

where D represents an average quantization error.

6. The method of claim 1, wherein said quantization matrix and said quantizer scale correspond substantially to coding parameters used in a coding operation that was previously performed on said encoded video data.

7. The method of claim 1, wherein said quantization matrix and said quantizer scale correspond substantially to coding parameters used in a coding operation that was previously performed on said encoded video data.

8. A method for evaluating the quality of encoded video data, the method comprising the steps of:

at least partially decompressing said encoded video data in an MPEG decoder and outputting a decompressed video data including a plurality of blocks;

detecting an intra-coded picture in each block of said decompressed video data;

if detected, performing a discrete cosine transform (DCT) on said decompressed video data to produce a set of DCT coefficients for at least one AC frequency band;

extracting quantization matrix data for the at least one AC frequency band;

extracting a quantizer scale for each block of said decompressed video data;

estimating a variance of said DCT coefficients;

determining an average quantization error for each set of said DCT coefficients; and

,

calculating a peak signal to noise ratio (PSNR) based on said average quantization error.

9. The method of claim 8, wherein said detecting step comprises the steps of:

performing a DC computation operation to recover the intra-dc-precision level from said decompressed video data; and

,

if said intra-dc-precision level is less than a predefined threshold, classifying said decoded video data as an intra-coded picture.

10. The method of claim 8, wherein said quantizer step size is calculated by averaging the AC coefficients in each of said decompressed video data as follows:

${\stackrel{\_}{C}}_{i,j}=\frac{C_{i,j}\times <br><br>16}{W_{i,j}},$

where C_{i, j} represents the (i, j)th AC coefficient in the current block, {overscore (C)}_{i, j} represents the normalized AC coefficient, and W_{i, j} represents the (i, j)th quantization matrix.

11. The method of claim 10, wherein the variance of the coefficients (λ

²_ij) is determined by the following equation;

${\mathrm{\lambda <br><br>}}_{i,j}^2=\frac{\underset{b=1}{\overset{N}{\sum <br><br>}}<br><br>{\left(C_{i,j}^b\right)}^2}{2<br><br>N}$

where C^b_ij represents the (i, j)th AC coefficient in each block (b), and N represents the total number of blocks.

12. The method of claim 11, wherein the average quantization error (D) is computed as follows:

$D=\frac{\underset{i=0,j=0}{\overset{I=7,j=7}{\sum <br><br>}}<br><br>D_{i,j}}{64},$

and D_{i, j}=2λ

²_{i, j},

where λ

²_ij represents the variance of the coefficients (λ

²_ij), and D_{i, j} represents the quantization error for the (i, j)th AC coefficient in each block.

13. The method of claim 12, wherein said PSNR is calculated as follows:

$\mathrm{PSNR}=10<br><br>\text{\hspace{1em}<br><br>}<br><br>{\log }_{10<br><br>\text{\hspace{1em}<br><br>}}<br><br>\frac{255\times <br><br>255}{D}$

where D represents an average quantization error.

14. An apparatus for evaluating the quality of encoded video data comprising:

a decoder for decoding at least a substantial portion of said encoded video data to produce decoded video data including a plurality of blocks;

a discrete cosine transform (DCT) configured to transform said decompressed video data into a set of DCT coefficients for at least one AC frequency band;

an extractor for extracting quantization matrix data for the at least one AC frequency band and for extracting a quantizer scale for each block of said decompressed video data;

a collector for estimating a variance of said DCT coefficients;

a first calculator for determining an average quantization error for each set of said DCT coefficients based on said variance, said quantization matrix, and said quantizer scale; and

,

a second calculator for determining a peak signal to noise ratio (PSNR) based on said average quantization error.

15. The apparatus of claim 14, further comprising a picture detector for detecting for an intra-coded picture in each block of said decompressed video data.

16. The apparatus of claim 15, wherein said picture detector comprises:

means for performing a DC computation operation to recover an intra-dc-precision level from said decompressed video data; and

,

means for classifying said decoded video data as an intra-coded picture when said intra-dc-precision level is less than a predefined threshold.

17. The apparatus of claim 14, wherein said quantization matrix and said quantizer scale correspond substantially to coding parameters used in a coding operation that was previously performed on said encoded video data.

18. The apparatus of claim 14, wherein said quantizer step size is calculated by averaging the AC coefficients in each of said decompressed video data as follows:

${\stackrel{\_}{C}}_{i,j}=\frac{C_{i,j}\times <br><br>16}{W_{i,j}},$

where C_{i, j} represents the (i, j)th AC coefficient in current block, {overscore (C)}_{i, j} represents the normalized AC coefficient, and W_{i, j} represents the (i, j)th quantization matrix.

19. The apparatus of claim 18, wherein the variance of the coefficients (λ

²_ij) is determined by the following equation;

${\mathrm{\lambda <br><br>}}_{i,j}^2=\frac{\underset{b=1}{\stackrel{N}{\sum <br><br>}}<br><br>{\left(C_{i,j}^b\right)}^2}{2<br><br>N},$

where C^b_ij represents the (i, j)th AC coefficient in each block (b), and N represents the total number of blocks.

20. The apparatus of claim 19, wherein the average quantization error (D) is computed as follows:

$D=\frac{\underset{i=0,j=0}{\stackrel{I=7,j=7}{\sum <br><br>}}<br><br>\text{\hspace{1em}<br><br>}<br><br>D_{i,j}}{64},$

and D_{i, j}=2λ

²_{i, j},

where λ

²_ij represents the variance of the coefficients (λ

²_ij), and D_{i, j} represents the quantization error for the (i, j)th AC coefficient in each block.

21. The apparatus of claim 20, wherein said PSNR is calculated as follows:

$\mathrm{PSNR}=10<br><br>\text{\hspace{1em}<br><br>}<br><br>{\log }_{10<br><br>\text{\hspace{1em}<br><br>}}<br><br>\frac{255\times <br><br>255}{D},$

where D represents an average quantization error.

22. An apparatus for evaluating the quality of encoded video data comprising:

a decoder configured to decode compressed variable-length Huffman codes and for producing therefrom decoded data, and for extracting quantization matrix data and a quantizer scale for each block of said decoded video data;

an inverse quantizer configured to perform inverse-quantizing of the decoded data output from said decoder to produce a set of DCT coefficients;

an inverse DCT configured to transform values of pixels in blocks of signals output from said inverse quantizer to dequantize decoded data including difference data;

a motion compensation and adder for receiving reference data within the encoded video data and said difference data from said inverse DCT to form motion-compensated pictures therefrom;

a collector coupled to the output of said inverse quantizer for estimating a variance of said DCT coefficients;

a first calculator for determining an average quantization error for each set of said DCT coefficients based on said variance, said quantization matrix, and said quantizer scale; and

,

a second calculator for determining a peak signal to noise ratio (PSNR) based on said average quantization error.

23. The apparatus of claim 22, further comprising a video memory configured to store reproduced video data.

24. The apparatus of claim 22, wherein said quantization matrix and said quantizer scale correspond substantially to coding parameters used in a coding operation that was previously performed on said encoded video data.

25. A system for evaluating the quality of encoded video data comprising:

a memory for storing a computer-readable code; and

,

a processor operatively coupled to said memory, said processor configured to;

decompress said encoded video data to produce a decompressed video data including a plurality of blocks;

detect an intra-coded picture in each block of said decompressed video data;

if detected, perform a discrete cosine transform (DCT) on said decompressed video data to produce a set of DCT coefficients for at least one AC frequency band;

extract quantization matrix data for the at least one AC frequency band;

extract a quantizer scale for each block of said decompressed video data;

estimate a variance of said DCT coefficients;

determine an average quantization error for each set of said DCT coefficients; and

,

calculate a peak signal to noise ratio (PSNR) based on said average quantization error.

26. The system of claim 25, wherein the detection of said intra-coded picture includes the steps of:

performing a DC computation operation to recover an intra-dc-precision level from said decompressed video data; and

,

classifying said decoded video data as an intra-coded picture when said intra-dc-precision level is less than a predefined threshold.

27. The apparatus of claim 25, wherein said quantization matrix and said quantizer scale correspond substantially to coding parameters used in a coding operation that was previously performed on said encoded video data.