System and method for quality-aware selection of parameters in transcoding of digital images

• US 8,559,739 B2
• Filed: 09/17/2012
• Issued: 10/15/2013
• Est. Priority Date: 12/03/2007
• Status: Active Grant

• Current Assignee: Ecole de technologie superieure
• Original Assignee: Ecole de technologie superieure
• Inventors: Coulombe, Stphane, Pigeon, Steven, Franche, Jean-Franois
• Primary Examiner(s): Zarka, David
• Assistant Examiner(s): LE, HA N
• Application Number:
  US13/621,329
• Publication Number:
  US 20130094778A1
• Time in Patent Office: 393 Days
• Field of Search: None
• US Class Current: 382/239
• CPC Class Codes: G06T 9/004 Predictors, e.g. intraframe, interframe coding
• CPC Class Codes: H04N 19/115 Selection of the code volume for a coding unit prior to coding
• CPC Class Codes: H04N 19/126 Details of normalisation or weighting functions, e.g. normalisation matrices or variable uniform quantisers
• CPC Class Codes: H04N 19/149 by estimating the code amount by means of a model, e.g. mathematical model or statistical model
• CPC Class Codes: H04N 19/154 Measured or subjectively estimated visual quality after decoding, e.g. measurement of distortion use of rate-distortion criteria H04N19/147
• CPC Class Codes: H04N 19/192 the adaptation method, adaptation tool or adaptation type being iterative or recursive
• CPC Class Codes: H04N 19/40 using video transcoding, i.e. partial or full decoding of a coded input stream followed by re-encoding of the decoded output stream

Abstract

A method for quality-aware transcoding of an input image into an output image for display on a terminal having device file size and image size constraints are disclosed. The method extracts features of the input image including dimensions and file size of the input image, predicts a file size of the output image taking into account constraints of the terminal and extracted features, including selecting a set of feasible transcoding parameters so that a corresponding predicted file size of the output image meets the device file size constraint of the terminal, determines quality metric values of the output image, characterizing a measure of distortion of the input image introduced by the transcoding, corresponding to the feasible transcoding parameters in the set of feasible transcoding parameters, and selects those transcoding parameters from the set of feasible transcoding parameters, which yield the highest quality metric value. A corresponding system is also provided.

Claims

1. A method for quality-aware transcoding of an input image into an output image for display on a terminal having device file size and image size constraints, the method comprising:

(a) extracting features of the input image including dimensions and a file size of the input image;

(b) predicting, from file size prediction information obtained by transcoding a set of training images, a file size of the output image taking into account the constraints of the terminal and the extracted features, comprising selecting a set of feasible transcoding parameters so that a corresponding predicted file size of the output image meets the device file size constraint of the terminal;

(c) determining, from quality metric prediction information obtained by transcoding the set of training images, predicted quality metric (QM) values of the output image, the predicted QM values characterizing a predicted measure of distortion of the input image introduced by the transcoding, corresponding to the feasible transcoding parameters in the set of feasible transcoding parameters;

the quality metric prediction information being determined explicitly by a comparison between input images from the set of training images and corresponding output images resulting from the transcoding of the set of training images; and

(d) selecting those transcoding parameters from the set of feasible transcoding parameters, which yield the highest predicted QM value, corresponding to the highest predicted visual quality for the output image for the set of feasible transcoding parameters.

2. The method of claim 1, further comprising indexing the file size prediction information by transcoding parameters including a transcoder scaling factor zT and an output encoding quality factor QFT of the output image.

3. The method of claim 2, further comprising indexing the quality metric prediction information by the transcoder scaling factor zT and the output encoding quality factor QFT.

4. The method of claim 2, wherein the step (b) further comprises obtaining the file size prediction information from an array of predicted relative output file sizes for image transcoding, obtained from the transcoding of the set of training images.

5. The method of claim 3, further comprising indexing the file size prediction information and the quality metric prediction information by an encoding quality factor QF(I) of the input image.

6. The method of claim 3, wherein the step (c) further comprises obtaining the quality metric prediction information from an array of predicted quality metric values for image transcoding, obtained from the transcoding of the set of training images.

7. The method of claim 1, wherein the step (c) comprises determining the quality metric values based on a Peak Signal to Noise Ratio (PSNR) measure or a Maximum Difference (MD) measure.

8. The method of claim 3, further comprising indexing the quality metric prediction information by a viewing scaling factor, characterizing a viewing condition of the output image.

9. A system for quality-aware transcoding of an input image into an output image for display on a terminal having device file size and image size constraints, the system comprising:

a processor;

a non-transitory computer readable storage medium having computer executable instructions stored thereon for execution by the processor, causing the processor to;

(a) extract features of the input image including dimensions and a file size of the input image;

(b) predict, from file size prediction information obtained by transcoding a set of training images, a file size of the output image taking into account the constraints of the terminal and the extracted features, comprising selecting a set of feasible transcoding parameters so that a corresponding predicted file size of the output image meets the device file size constraint of the terminal;

(c) determine, from quality metric prediction information obtained by transcoding the set of training images, predicted quality metric (QM) values of the output image, characterizing a predicted measure of distortion of the input image introduced by the transcoding, corresponding to the feasible transcoding parameters in the set of feasible transcoding parameters;

the quality metric prediction information being determined explicitly by a comparison between input images from the set of training images and corresponding output images resulting from the transcoding of the set of training images; and

(d) select those transcoding parameters from the set of feasible transcoding parameters, which yield the highest predicted QM value, corresponding to the highest predicted visual quality for the output image for the set of feasible transcoding parameters.

10. The system of claim 9, wherein the file size prediction information is indexed by transcoding parameters including a transcoder scaling factor zT and an output encoding quality factor QFT of the output image.

11. The system of claim 10, wherein the quality metric prediction information is indexed by the transcoder scaling factor zT and the output encoding quality factor QFT.

12. The system of claim 10, wherein the file size prediction information comprises an array of predicted relative output file sizes, obtained from transcoding of the set of training images.

13. The system of claim 11, wherein the file size prediction information and the quality metric values prediction information are further indexed by an encoding quality factor QF(I) of the input image.

14. The system of claim 11, wherein the quality metric prediction information comprises an array of predicted quality metric values for image transcoding, obtained from the transcoding of the set of training images.

15. The system of claim 9, wherein the computer readable instructions further cause the processor to determine the quality metric values based on a Peak Signal to Noise Ratio (PSNR) measure or a Maximum Difference (MD) measure.

16. The system of claim 11, wherein the quality metric prediction information is further indexed by a viewing scaling factor, characterizing a viewing condition of the output image.

17. A system for quality-aware transcoding of an input image into an output image for display on a terminal having device file size and image size constraints, the system comprising:

a processor; and

a non-transitory computer readable storage medium having computer executable instructions stored thereon for execution by the processor, forming;

(a) a means for extracting features of the input image including dimensions and a file size of the input image;

(b) a means for predicting, from file size prediction information obtained by transcoding of a set of training images, a file size of the output image taking into account the constraints of the terminal and the extracted features, comprising selecting a set of feasible transcoding parameters so that a corresponding predicted file size of the output image meets the device file size constraint of the terminal;

(c) a means for determining, from quality metric prediction information obtained by transcoding of the set of training images, predicted quality metric (QM) values of the output image, the predicted QM values characterizing a predicted measure of distortion of the input image introduced by the transcoding, corresponding to the feasible transcoding parameters in the set of feasible transcoding parameters;

the quality metric prediction information being determined explicitly by a comparison between input images from the set of training images and corresponding output images resulting from the transcoding of the set of training images; and

(d) a means for selecting those transcoding parameters from the set of feasible transcoding parameters, which yield the highest predicted QM value, corresponding to the highest predicted visual quality for the output image for the set of feasible transcoding parameters.

18. The system of claim 17, wherein the means (c) are configured to determine the quality metric values based on a Peak Signal to Noise Ratio (PSNR) measure or a Maximum Difference (MD) measure.

19. The system of claim 17, wherein the file size prediction information is indexed by transcoding parameters including a transcoder scaling factor zT and an output encoding quality factor QFT of the output image.

20. The system of claim 19, wherein the quality metric prediction information is indexed by the transcoder scaling factor zT and the output encoding quality factor QFT.

21. The system of claim 20, wherein the file size prediction information and the quality metric prediction information respectively comprise an array of predicted relative output file sizes and an array of predicted quality metric values, obtained from the transcoding of the set of training images.

22. The system of claim 20, wherein the file size prediction information and the quality metric prediction information are further indexed by one of the following:

an encoding quality factor QF(I) of the input image;

or

a viewing scaling factor, characterizing a viewing condition of the output image.