Classifying an image's compression level

• US 8,452,110 B2
• Filed: 03/31/2011
• Issued: 05/28/2013
• Est. Priority Date: 10/05/2009
• Status: Active Grant

• Current Assignee: Beamr Imaging Ltd.
• Original Assignee: I.C.V.T Ltd.
• Inventors: Carmel, Sharon, Gill, Dror, Shoham, Tamar
• Primary Examiner(s): COUSO, JOSE L
• Application Number:
  US13/077,060
• Publication Number:
  US 20110222786A1
• Time in Patent Office: 789 Days
• Field of Search: 382232-233, 382244-251, 358/426.14, 348/387.1, 348/390.1, 348/395.1, 348/403.1, 37524018-2402, 375/E7.144, 375/E7.154
• US Class Current: 382/232
• CPC Class Codes: H04N 19/124 Quantisation
• CPC Class Codes: H04N 19/126 Details of normalisation or weighting functions, e.g. normalisation matrices or variable uniform quantisers
• CPC Class Codes: H04N 19/136 Incoming video signal characteristics or properties
• CPC Class Codes: H04N 19/14 Coding unit complexity, e.g. amount of activity or edge presence estimation H04N19/146 takes precedence
• 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 19/17 the unit being an image region, e.g. an object
• CPC Class Codes: H04N 19/176 the region being a block, e.g. a macroblock
• CPC Class Codes: H04N 19/182 the unit being a pixel
• CPC Class Codes: H04N 19/192 the adaptation method, adaptation tool or adaptation type being iterative or recursive
• CPC Class Codes: H04N 19/50 using predictive coding H04N19/61 takes precedence
• CPC Class Codes: H04N 19/60 using transform coding
• CPC Class Codes: H04N 19/625 using discrete cosine transform [DCT]
• CPC Class Codes: H04N 19/86 involving reduction of coding artifacts, e.g. of blockiness

Abstract

A method of classifying a compressed input image'"'"'s compression level, comprising applying inverse quantization to DCT coefficients extracted from the compressed input image, giving rise to dequantized DCT coefficients, applying a threshold to the dequantized DCT coefficients, and determining a compression level classification for the compressed input image according to a relation among the threshold and at least one of the dequantized DCT coefficients.

Claims

1. A program storage device readable by machine, tangibly embodying a program of instructions executable by the machine to perform a method of classifying a compressed input image'"'"'s compression level, comprising:

applying inverse quantization to discrete cosine transform (DCT) coefficients extracted from the compressed input image giving rise to dequantized DCT coefficients;

identifying a lowest absolute non-zero DCT coefficient among said dequantized DCT coefficients;

applying a threshold to said dequantized DCT coefficients,

wherein said applying a threshold includes applying a threshold to said lowest absolute non-zero DCT coefficient; and

determining a compression level classification for said compressed input image according to a relation among said threshold and at least one of said dequantized DCT coefficients,

wherein said determining a compression level classification, comprises determining a compression level classification for said compressed image according to a relation among said threshold and said lowest absolute non-zero DCT coefficient.

2. The method according to claim 1, wherein applying a threshold comprises applying a plurality of thresholds which relate to different characteristics or components of the compressed input image.

3. The method according to claim 1, wherein the threshold includes a threshold that is related to DCT coefficients in the Luma (Y) plane.

4. The method according to claim 1, wherein the plurality of thresholds include a threshold that is related to DCT coefficients in the Luma (Y) plane of the compressed input image, and further include a threshold or thresholds that is/are related to DCT coefficients in each of the Chroma planes (U and V).

5. The method according to claim 4, wherein said applying a threshold comprises applying a first threshold to said dequantized DCT coefficients in the Luma (Y) plane, and said determining a compression level comprises classifying said image'"'"'s compression level as high if its lowest absolute non-zero DCT value in the Luma (Y) plane is above the first threshold.

6. The method according to claim 5, wherein said applying a threshold comprises applying a second threshold to said dequantized DCT coefficients in the Luma (Y) plane, and said determining a compression level comprises classifying said image'"'"'s compression level as low if its lowest absolute non-zero DCT value in the Luma (Y) plane is below the second threshold.

7. The method according to claim 6, wherein in case the image'"'"'s lowest absolute non-zero DCT value in the Luma (Y) plane is not above said first threshold or below said second threshold, said applying a threshold comprises applying a third threshold and a fourth threshold to dequantized DCT coefficients in the Chroma U plane and to dequantized DCT coefficients in the Chroma V plane, respectively, and said determining a compression level comprises classifying said image'"'"'s compression level as high if its lowest absolute non-zero DCT value in the Chroma U plane is above the respective third threshold and its lowest absolute non-zero DCT value in the Chroma V plane is above the respective fourth threshold.

8. The method according to claim 7, wherein said third and fourth thresholds are identical.

9. The method according to claim 6, wherein in case the image'"'"'s lowest absolute non-zero DCT value in the Luma (Y) plane is not above said first threshold or below said second threshold, said applying a threshold comprises applying a third threshold and a fourth threshold to dequantized DCT coefficients in the Chroma U plane and to dequantized DCT coefficients in the Chroma V plane, respectively, and said determining a compression level comprises classifying said image'"'"'s compression level as low if its lowest absolute non-zero DCT value in the Chroma U plane is below the respective third threshold or its lowest absolute non-zero DCT value in the Chroma V plane is below the respective fourth threshold.

10. The method according to claim 1, wherein said threshold is related to a perceptual similarity criterion and to a further compression criterion.

11. The method according to claim 10, wherein said threshold is characterized in terms of estimated potential for further compression of said image given a perceptual similarity constraint, the perceptual similarity constraint setting a limit on an extent of perceptual similarity loss as a result of further compression.

12. The method according to claim 11, wherein the extent of perceptual similarity loss as a result of further compression is limited to a degree which maintains perceptual identify between the image before compression and the image after compression.

13. The method according to claim 11, wherein according to said threshold a compression of said image is regarded as being high, if it is estimated that, given the perceptual similarity loss threshold, less than a minimum compression ratio is achievable through compression of said image.

14. The method according to claim 11, wherein according to said threshold a compression of said image is regarded as being low, if it is estimated that, given the perceptual similarity loss threshold, more than a minimum compression ratio is achievable through compression of said image.

15. The method according to claim 10, wherein said perceptual similarity criterion is associated with a specific perceptual similarity measure or with a combination of specific perceptual similarity measures and with specific configurations thereof.

16. The method according to claim 15, wherein said threshold relates to a specific compression technique or algorithm and specific configurations thereof, and wherein said compressed input image was compressed using corresponding compression technique or algorithm and configurations thereof.

17. The method according to claim 11, wherein said threshold relates to a DCT coefficient value, and a compression of said image is regarded as high if the image'"'"'s lowest absolute non-zero DCT value is above the DCT coefficient value set forth by the threshold.

18. The method according to claim 1, further comprising determining a threshold by empirically evaluating statistics over a plurality of recompressed images that were compressed to a maximum extent that is allowed by a perceptual similarity constraint, the perceptual similarity constraint setting a limit on an extent of perceptual similarity loss as a result of further compression.

19. The method according to claim 18, wherein said determining a threshold comprises identifying images for which the substantially maximal compression ratio that is achievable by recompressing said images under the perceptual similarity constraint, is below a predefined further compression ratio.

20. The method according to claim 19, wherein said determining a threshold comprises identifying, based on said plurality of images, a minimum lowest absolute non-zero DCT coefficient which characterizes highly compressed images, being the smallest lowest absolute non-zero DCT value of an image, for which a maximal compression ratio that is achievable by recompression under the perceptual similarity constraint is always, based on said plurality of images, below the predefined further compression ratio.

21. The method according to claim 20, wherein the threshold which characterizes highly compressed images is a first threshold, and wherein said determining a threshold further comprises determining a second threshold, and wherein said determining a second threshold comprises identifying images for which the substantially maximal compression ratio that is achievable by recompressing said images under the perceptual similarity constraint, is above the predefined further compression ratio.

22. The method according to claim 21, wherein said determining a second threshold comprises identifying, based on said plurality of images, a maximum lowest absolute non-zero DCT coefficient which characterizes highly compressed images, being the largest lowest absolute non-zero DCT value of an image, for which a maximal compression ratio that is achievable by recompression under the perceptual similarity constraint is always, based on said plurality of images, above the predefined further compression ratio.

23. The method according to claim 22, wherein an intermediate range of lowest absolute non-zero DCT coefficient is defined between the minimum lowest absolute non-zero DCT coefficient associated with the first threshold and the maximum lowest absolute non-zero DCT coefficient associated with the second threshold.

24. The method according to claim 23, wherein the minimum lowest absolute non-zero DCT coefficient associated with the first threshold and the maximum lowest absolute non-zero DCT coefficient associated with the second threshold are both in the Y plane of an image, and wherein said determining a threshold further comprises determining a third threshold for classifying images whose lowest absolute non-zero DCT coefficient falls within the intermediate range, and wherein said determining a third threshold comprises identifying, based on said plurality of images, a minimum lowest absolute non-zero DCT coefficient in the U and V planes which characterizes highly compressed images, being the smallest lowest absolute non-zero DCT value in the U and V planes of an image, for which a maximal compression ratio that is achievable by recompression under the perceptual similarity constraint is always, based on said plurality of images, below the predefined further compression ratio.

25. The method according to claim 24, wherein if a lowest absolute non-zero DCT coefficient in the Y plane of the compressed input image is the intermediate range, said determining a compression level classification comprises, classifying a compressed input image as being highly compressed if the lowest absolute non-zero DCT coefficient in the U and V planes are both above the third threshold.

26. The method according to claim 24, wherein said third threshold comprises a different threshold for DCT coefficients in each of the U and V planes.

27. The method according to claim 1, wherein said threshold is sensitive to a certain characteristic or to a certain component of the compressed input image and wherein said applying a threshold to said dequantized DCT coefficients comprises adapting said threshold according to a certain characteristic or to a certain component of the compressed input image.

28. The method according to claim 27, wherein said threshold is sensitive to one or more of the following:

a spatial resolution of the compressed input image, the variability of DCT coefficients between neighboring blocks of the compressed input image and the frequency content of the compressed input image.

29. A computer program product comprising a computer useable medium having computer readable program code embodied therein of classifying a compressed input image'"'"'s compression level, the computer program product comprising:

computer readable program code for causing the computer to apply inverse quantization to discrete cosine transform (DCT) coefficients extracted from the compressed input image giving rise to dequantized DCT coefficients,

computer readable program code for causing the computer to identify a lowest absolute non-zero DCT coefficient among said dequantized DCT coefficients;

computer readable program code for causing the computer to apply a threshold to said dequantized DCT coefficients,

the computer readable program code for causing the computer to apply a threshold including computer readable program code for causing the computer to apply a threshold to said lowest absolute non-zero DCT coefficient; and

computer readable program code for causing the computer to determine a compression level classification for said compressed input image according to a relation among said threshold and at least one of said dequantized DCT coefficients,

the computer readable program code for causing the computer to determine a compression level classification including computer readable program code for causing the computer to determining a compression level classification for said compressed image according to a relation among said threshold and said lowest absolute non-zero DCT coefficient.

30. A computerized apparatus for classifying a compressed input image'"'"'s compression level, comprising:

an inverse quantizer adapted to apply inverse quantization to discrete cosine transform (DCT) coefficients extracted from the compressed input image giving rise to dequantized DCT coefficients;

a configuration module adapted to provide a classification threshold associated; and

a DCT-domain classifier adapted to identify a lowest absolute non-zero DCT coefficient among said dequantized DCT coefficients, adapted to apply the classification threshold to the dequantized DCT coefficients including applying the classification threshold to said lowest absolute non-zero DCT coefficient, and adapted to determine a compression level classification for the compressed input image according to a relation among the classification threshold and at least one of the dequantized DCT coefficients including determining the compression level classification for said compressed image according to a relation among said threshold and said lowest absolute non-zero DCT coefficient.