System and method for enhancing data compression using dynamic learning and control

Patent metadata

Field	Value
Publication number	US-9565439-B2
Application number	US-57947409-A
Country	US
Kind code	B2
Filing date	Oct 15, 2009
Priority date	Oct 15, 2009
Publication date	Feb 7, 2017
Grant date	Feb 7, 2017

Abstract

A data compression system and method for achieving enhanced compression performance in a plurality of encoder-decoder pairs communicatively coupled in a concatenated chain is disclosed. The plurality of encoder-decoder pairs receive input data and transform the input data into compressed output data. The data compression system includes at least one quality estimator communicatively coupled to the concatenated chain to determine a plurality of quality metrics corresponding to an output from each of the plurality of encoder-decoder pairs. Further, the data compression system also includes a processing subsystem configured to optimize the operation of each of the plurality of encoder-decoder pairs based upon the plurality of quality metrics to achieve a desired level of optimization for the concatenated chain.

First claim

The invention claimed is:

1. A data compression system, comprising:
a plurality of encoder-decoder pairs communicatively coupled in a concatenated chain and configured to receive input data and transform the input data into compressed output data, wherein the plurality of encoder-decoder pairs each comprise:
a first encoder-decoder pair comprising:
a first encoder configured to compress and to encode the input data; and
a first decoder configured to decode the data encoded and compressed by the first encoder and to provide decoded data to a subsequent encoder-decoder pair; and
at least one additional encoder-decoder pair comprising:
an additional encoder configured to receive decoded data from a prior encoder-decoder pair and to compress and encode the decoded data; and
an additional decoder configured to decode the data encoded by the additional encoder and to provide the data decoded by the additional decoder to a subsequent encoder-decoder pair or provide the data decoded by the additional decoder as output data of the concatenated chain when there are no subsequent encoder-decoder pairs in the concatenated chain;

at least one quality estimator communicatively coupled to the concatenated chain to determine one or more quality metrics corresponding to the decoded data provided by the first decoder, the decoded data provided by the additional decoder of the at least one addition encoder-decoder pairs, or a combination thereof; and
a processing subsystem configured to optimize operation of each of the plurality of encoder-decoder pairs based upon the plurality of quality metrics, to achieve a desired level of optimization for the concatenated chain.

2. The data compression system of claim 1 , further comprising a plurality of quality estimators, wherein each of the plurality of quality estimators is communicatively coupled to a corresponding encoder-decoder pair.

3. The data compression system of claim 2 , wherein each of the plurality of quality estimators generates a quality metric for a corresponding encoder-decoder pair.

4. The data compression system of claim 1 , wherein the processing subsystem optimizes each of the plurality of encoder-decoder pairs based on a corresponding quality metric.

5. The data compression system of claim 1 , wherein the plurality of quality metrics comprise a just noticeable difference, a peak signal-to-noise ratio, a structural similarity, a compression rate, an image quality, an audio quality, or combinations thereof.

6. The data compression system of claim 1 , wherein the processing subsystem optimizes each of the plurality of encoder-decoder pairs based on content of the input data.

7. The data compression system of claim 6 , wherein the processing subsystem optimizes each of the plurality of encoder-decoder pairs based upon an object size, a location, a velocity vector, a complexity of motion, or combinations thereof.

8. The data compression system of claim 1 , wherein the input data comprises image data, audio data or a combination of image data and audio data.

9. The data compression system of claim 8 , wherein the processing subsystem analyzes the image data, the audio data or the combination of the image data and the audio data in an image domain, a transform domain, or a combination thereof.

10. The data compression system of claim of claim 1 , wherein the processing subsystem is configured to employ a genetic algorithm, a simulated annealing technique, a hill climbing method, fuzzy logic, and combinations thereof, to optimize operation of each of the plurality of encoder-decoder pairs.

11. The data compression system of claim 1 , wherein the processing subsystem is configured to iteratively vary one or more parameters corresponding to the plurality of encoder-decoder pairs to determine optimal values that achieve a desired quality or compression rate or combined quality and compression rate of the output data.

12. The data compression system of claim 11 , further comprising a controller that updates the one or more parameters corresponding to the plurality of encoder-decoder pairs with the determined optimal values.

13. The data compression system of claim 12 , wherein the controller updates the one or more parameters corresponding to the plurality of encoder-decoder pairs with the optimal values at determined time intervals, or substantially in real-time.

14. The data compression system of claim 11 , further comprising a data repository for storing at least one of the input data, the plurality of quality metrics corresponding to an output from each of the plurality of encoder-decoder pairs, and the optimal values corresponding to the one or more parameters corresponding to the plurality of encoder-decoder pairs.

15. The data compression system of claim 1 , wherein the processing subsystem comprises at least one of a linear classifier, a perceptron, a decision tree, a neural network, a support vector machine, perpetual quartz systems, and a field programmable gate array, to optimize operation of each of the plurality of encoder-decoder pairs for achieving a desired level of optimization for the concatenated chain.

16. The data compression system of claim 15 , wherein the neural network is a physiological analog of the human visual cortex.

17. The data compression system of claim 1 , wherein the at least one quality estimator determines the quality metrics based on a comparison of a quality metric associated with output from an encoder-decoder pair and a corresponding quality metric associated with a reference data, or based on an estimation function for determining quality metrics corresponding to the output from the encoder-decoder pair.

18. A method for improving data compression, comprising:
identifying one or more desired characteristics of input data provided to a plurality of encoder-decoder pairs communicatively coupled in a concatenated chain, the plurality of encoder-decoder pairs configured to receive the input data and transform the input data into compressed output data by:
using a first encoder-decoder pair to:
compress and encode the input data, creating compressed and encoded input data, decode the compressed and encoded input data, creating decoded input data, and provide the decoded input data to a subsequent encoder-decoder pair; and
using at least one additional encoder-decoder pair to:
receive a decoded output from a previous encoder-decoder pair,
compress and encode the decoded output from the previous encoder-decoder pair, creating subsequent compressed and encoded data,
decode the subsequent compressed and encoded data, creating subsequent decoded data; and
provide the subsequent decoded data to a subsequent additional encoder-decoder pair of the concatenated chain or provide the subsequent decoded data as output data of the concatenated chain when there are no subsequent encoder-decoder pairs in the concatenated chain;
estimating one or more quality metrics corresponding to an output from each of the plurality of encoder-decoder pairs; and
optimizing operation of each of the plurality of encoder-decoder pairs based upon the plurality of quality metrics, to achieve a desired level of optimization for the concatenated chain.

19. The method of claim 18 , wherein optimizing operation of each of the plurality of encoder-decoder pairs comprises iteratively varying one or more parameters corresponding to one or more of the encoder-decoder pairs to determine optimal values that achieve a desired output quality or compression rate or combined output quality and compr

Assignees

• Bush Stephen Francis
• Hershey John Erik
• Dell'Anno Michael J
• Nbcuniversal Media Llc

Inventors

• Bush Stephen Francis
• Hershey John Erik
• Dell'Anno Michael J

Classifications

• H04N19/196

  being specially adapted for the computation of encoding parameters, e.g. by averaging previously computed encoding parameters (processing of motion vectors H04N19/513) · CPC title

• H04N19/15

  by monitoring actual compressed data size at the memory before deciding storage at the transmission buffer · CPC title

• H04N19/154

  Measured or subjectively estimated visual quality after decoding, e.g. measurement of distortion (use of rate-distortion criteria H04N19/147) · CPC title

• H04N19/12 Primary

  Selection from among a plurality of transforms or standards, e.g. selection between discrete cosine transform [DCT] and sub-band transform or selection between H.263 and H.264 · CPC title

• H04N19/436

  using parallelised computational arrangements · CPC title