Geometric vector quantization

• US 5,371,544 A
• Filed: 02/07/1992
• Issued: 12/06/1994
• Est. Priority Date: 02/07/1992
• Status: Expired due to Term

• Current Assignee: American Telephone & Telegraph Company (AT&T, Inc.)
• Original Assignee: AT&T Corporation (AT&T, Inc.)
• Inventors: Jayant, Nuggehally S., Jacquin, Arnaud E., Podilchuk, Christine I.
• Primary Examiner(s): Kostak, Victor R.
• Application Number:
  US07/832,536
• Time in Patent Office: 1,033 Days
• Field of Search: 358/133, 358/135, 358/136, 358/138, 358/426, 358/433, 382/56, 341/106, 375/27, 375/122, 381/29, 381/30, 381/31, 348/398, 348/422
• US Class Current: 375/240.11
• CPC Class Codes: G06T 9/008 Vector quantisation
• CPC Class Codes: H03M 7/3082 Vector coding for television signals, see H04N19/94
• CPC Class Codes: H04N 19/30 using hierarchical techniques, e.g. scalability H04N19/63 takes precedence
• CPC Class Codes: H04N 19/503 involving temporal prediction adaptive coding with adaptive selection between spatial and temporal predictive coding H04N19/107; adaptive coding with adaptive selection among a plurality of temporal predictive coding modes H04N19/109
• CPC Class Codes: H04N 19/62 by frequency transforming in three dimensions H04N19/63 takes precedence
• CPC Class Codes: H04N 19/94 Vector quantisation

Abstract

A geometric vector quantizer coding technique is illustrated in the context of a full motion video coder based on a three-dimensional sub-band framework. The original image data is decomposed into different spatial-temporal frequency bands and based on the data in each band, different quantization strategies are applied to the bands. The lowest spatial-temporal frequency band is coded using a traditional three-dimensional switched predictor and optimum scaler quantizer. The non-dominant sub-bands are coded using the vector quantization approach to efficiently encode the images while appropriately exploiting the sparse, highly structured nature of the data to design the codebooks. Variable resolution is achieved using these techniques and no training is needed to establish or maintain the codebook. A fast codebook search technique is provided for identifying the optimal codebook vector for representing a block of input data. Examples of two and three level geometric vector quantizers are also provided.

Claims

1. A method of coding a set of multi-level signals comprising the steps of:

ordering the multi-level signals according to values of a signal parameter;

one or more times, partitioning the order of multi-level signals into two or more elements, each element including one or more multi-level signals;

determining a magnitude for an element based upon signal parameter values for signals in the element;

comparing a partition of elements with the set of multi-level signals to determine a difference metric based on partition element magnitudes and multi-level signal parameter values; and

selecting as a vector representing the set of multi-level signals a partition of elements which satisfies a difference metric criterion.

2. The method of claim 1 wherein the signal parameter comprises signal intensity.

3. The method of claim 1 wherein the magnitude of a partition element is determined based on an average signal parameter value for one or more multi-level signals of the element.

4. The method of claim 3 wherein the signal parameter value comprises signal intensity value.

5. The method of claim 1 wherein the step of determining a magnitude comprises the step of determining a magnitude for each element of a partition.

6. The method of claim 1 wherein the difference metric comprises a mean square difference.

7. The method of claim 1 wherein the difference metric criterion comprises a minimum difference metric.

8. The method of claim 1 wherein the step of selecting comprises determining a codebook index to represent the selected vector.

9. The method of claim 8 further comprising the step of communicating the determined codebook index to a receiver.

10. The method of claim 1 wherein the set of multi-level signals comprises a set of image pixels.

11. The method of claim 10 wherein a signal level reflects a gray level of a pixel.

12. A method of coding a set of multi-level signals comprising the steps of:

providing a set of vectors, each of said vectors comprising a geometric array of elements, wherein one or more first elements of each vector have associated therewith predetermined magnitudes, and wherein one or more other elements of each vector have associated therewith magnitudes determined based on levels of multi-level signals being coded;

comparing multi-level signals with one or more of the vectors to determine, for each such vector, a difference metric based on multi-level signal levels and vector element magnitudes; and

representing the set of multi-level signals with use of a vector which satisfies a difference metric criterion.

13. The method of claim 12 wherein the first difference metric criterion comprises a minimum difference metric.

14. The method of claim 12 wherein the step of representing the set of multi-level signals comprises the step of determining a codebook index to represent the vector which satisfies the first difference metric criterion and further comprising the step of communicating the determined codebook index to a receiver.

15. The method of claim 12 wherein the step of providing a set of vectors comprises the step of selecting one or more first elements of a vector to have predetermined magnitudes, the selection of the first elements based on the multi-level signals.

16. A method of coding a set of multi-level signals, wherein the set of multi-level signals comprises at least one signal having a positive level and at least one signal having a negative level, the method comprising the steps of:

a. ordering multi-level signals of the set according to signal level;

b. a plurality of times, partitioning the ordered, non-positive multi-level signals into at least two groups of non-positive signals;

c. for a partition of ordered, non-positive multi-level signals,

1. forming a first vector of signals, the first vector including first and second sets of signals,

i. the signals of the first vector'"'"'s first set having predetermined magnitudes and corresponding to a first group of non-positive multi-level signals in the partition, and

ii. the signals of the first vector'"'"'s second set having magnitudes based on levels of signals in a second group of non-positive multi-level signals in the partition;

2. comparing the first vector of signals with ordered, non-positive multi-level signals to determine a first difference metric;

d. identifying a first vector of signals having a first difference metric which satisfies a first difference metric criterion;

e. a plurality of times, partitioning the ordered, non-negative multi-level signals into at least two groups of non-negative signals;

f. for a partition of ordered, non-negative multi-level signals,

18. The method of claim 16 wherein each of the first and second difference metric criteria comprises a minimum difference metric.

21. The method of claim 18 wherein each of the first 2 and second difference metric criteria comprises a minimum difference metric.

17. forming a second vector of signals, the second vector including first and second sets of signals,

i. the signals of the second vector'"'"'s first set having predetermined magnitudes and corresponding to a first group of non-negative multi-level signals in the partition, and

ii. the signals of the second vector'"'"'s second set having magnitudes based on levels of signals in a second group of non-negative multi-level signals in the partition;

2. comparing the second vector of signals with ordered, non-negative multi-level signals to determine a second difference metric;

g. identifying a second vector of signals having a difference metric which satisfies a second difference metric criterion; and

h. combining the identified first and second vectors of non-positive and non-negative signals to determine a vector of signals for use in coding the set of multi-level signals.

19. A method of coding a set of multi-level signals, wherein the set of multi-level signals includes non-positive signal levels only, the method comprising the steps of:

a. ordering multi-level signals of the set according to signal level;

b. a plurality of times, forming a vector of signals by

2. partitioning the ordered, multi-level signals into first and second groups of signals, the first group of signals comprising at least one of the least negative multi-level signals and the second group comprising one or more of the balance of the multi-level signals,

2. assigning a predetermined magnitude to one or more first signals of the vector, which first signals correspond to the signals of the first group of signals,

20. a plurality of times, partitioning the second group of multi-level signals into first and second sub-groups of multi-level signals,for a partition of the second group of multi-level signals

i. assigning a magnitude to one or more second signals of the vector based on the multi-level signals of the first sub-group of multi-level signals,

ii. assigning a magnitude to one or more third signals of the vector based on the multi-level signals of the second sub-group of multi-level signals,

iii. comparing the second and third signals of the vector with ordered, non-positive multi-level signals to determine a first difference metric;

5. identifying second and third signals of the vector having a .first difference metric which satisfies a first difference metric criterion; and

6. comparing the vector of signals with ordered, multi-level signals to determine a second difference metric based on vector signal magnitudes and multi-level signal levels; and

c. identifying a vector of signals for use in coding the multi-level signals, the identified vector having a second difference metric which satisfies a second difference metric criterion.

23. The method of claim 20 wherein each of the first and second difference metric criteria comprises a minimum difference metric.

22. A method of coding a set of multi-level signals, wherein the set of multi-level signals includes non-negative signal levels only, the method comprising the steps of:

a. ordering multi-level signals of the set according to signal level;

b. a plurality of times, forming a vector of signals by

1. partitioning the ordered, multi-level signals into first and second groups of signals, the first group of signals comprising at least one of the least positive multi-level signals and the second group comprising one or more of the balance of the multi-level signals,

2. assigning a predetermined magnitude to one or more first signals of the vector, which first signals correspond to the signals of the first group of signals,

3. a plurality of times, partitioning the second group of multi-level signals into first and second sub-groups of multi-level signals,

4. for a partition of the second group of multi-level signals,

i. assigning a magnitude to one or more second signals of the vector based on the multi-level signals of the first sub-group of multi-level signals,

ii. assigning a magnitude to one or more third signals of the vector based on the multi-level signals of the second sub-group of multi-level signals;

iii. comparing the second and third signals of the vector with ordered, non-negative multi-level signals to determine a first difference metric;

5. identifying second and third signals of the vector having a first difference metric which satisfies a first difference metric criterion;

6. comparing the vector of signals with ordered, multi-level signals to determine a second difference metric based on vector signal magnitudes and multi-level signal levels; and

c. identifying a vector of signals for use in coding the multi-level signals, the identified vector having a second difference metric which satisfies a second difference metric criterion.