Method and apparatus for vector quantization

• US 5,067,152 A
• Filed: 03/16/1990
• Issued: 11/19/1991
• Est. Priority Date: 01/30/1989
• Status: Expired due to Fees

• Current Assignee: Medi-Tech
• Original Assignee: Medi-Tech
• Inventors: Kisor, Gregory H., King, Edward J.
• Primary Examiner(s): Tarcza, Thomas H.
• Assistant Examiner(s): Swann, Tod
• Application Number:
  US07/494,658
• Time in Patent Office: 613 Days
• Field of Search: 380/10, 358/133, 358/135, 358/136, 358/11, 358/12, 358/432, 358/261.3, 341/106, 325/27, 325/122, 381/31, 381/32
• US Class Current: 348/422.1
• 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 11/042 Codec means
• CPC Class Codes: H04N 19/94 Vector quantisation

Abstract

A method of and apparatus for vector quantizaiton operates on binary signal vectors. Each signal vector contains a plurality of sample elements of a signal to be compressed. Each of the signal vectors is hashed to generate a value identifying one of at least three different subsets of code words collectively constituting a set of binary code words. Each of the signal vectors is hashed a second time to generate a value identifying one of at least three different subsets of code words collectively constituting a second different set of binary code words. Each code word is equal in length to the length of each of the signal vectors and has associated with it a unique binary code of a lesser length. The vector quantizer performs a distortion comparison search through the code words of the selected subset to identify the one code word and its uniquely associated code which has the least distortion of all of the code words of the one subset with respect to the hashed signal vector. That one code is outputted to represent the hashed signal vector. The subsets may be mutually exclusive and include a plurality of code words.

Claims

1. A method of compressing an original signal convertible into a plurality of binary signal vectors, each signal vector containing a plurality of data elements of the original signal to be compressed, the method comprising the steps of:

hashing a first signal vector converted from the original signal to identify one of at least three different subsets of a first set of code words, each code word of the first set having a unique, associated binary code, each code associated with the first set of code words having a first predetermined length less than the length of the first signal vector;

storing a first code associated with one of the code words from the one identified subset of the first set as a compressed representation of the first signal vector;

hashing a second signal vector converted from the original signal to identify one of at least three different subsets of a second set of code words, each code word of the second set having a unique, associated binary code, each code associated with the second set of code words having a second predetermined length less than the length of the second signal vector and the second set of code words differing from the first set of code words; and

storing a second code associated with one of the code words from the one selected subset of the second set as a compressed representation of the second signal vector.

2. The method of claim 1 further comprising converting the original signal to be compressed into the plurality of signal vectors including the first signal vector and the second signal vector.

3. The method of claim 2 wherein the original signal to be compressed is a composite signal and wherein each data element of the composite signal is a composite of plural component data elements wherein the converting step comprises the steps of:

converting composite data elements from a portion of the composite signal into the first signal vector; and

converting at least some of the same composite data elements from the portion of the composite signal into the second signal vector.

4. The method of claim 1 wherein the original signal to be compressed comprises a video signal and wherein one of the first and second signal vectors contains luminance data from the video signal.

5. The method of claim 4 wherein the video signal is a color video signal and wherein the remaining one of the first and second signal vectors contains chrominance data from the color video signal.

6. The method of claim 5 wherein the remaining one of the first and second signal vectors contains hue data from the color video signal and further comprising the steps of:

hashing a third signal vector converted from the original signal to identify one subset of a third set of code words, the third signal vector containing saturation data from the color video signal, the third set of code words differing from each of the first and second sets of code words, each code word of the third set having a unique, associated binary code, each code associated with the third set being less than the length of the third signal vector; and

storing a code associated with a code word of the one identified subset of the third set of code words as a compressed representation of the third signal vector.

7. The method of claim 1 further comprising the collective steps of hashing each of a remaining plurality of signal vectors converted from the signal, identifying a code associated with a code word of one of a plurality of different sets of code words, including the first and second sets, for each of the hashed signal vectors, and storing the plurality of identified codes, including the first and second codes, on a record storage element as a compressed representation of the original signal.

8. A record storage medium storing the plurality of codes generated in accordance with the steps of method claim 7 as a compressed representation of the original signal.

9. The method of claim 1 further comprising the collective steps of having each of a remaining plurality of signal vectors converted from the original signal, identifying a code associated with a code word of one of a plurality of different sets of code words, including the first and second sets, for each of the hashed signal vectors, and transmitting the plurality of identified codes, including the first and second codes, as a compressed binary signal representation of the data elements of the original signal.

10. The binary signal generated in accordance with the steps of the method of claim 9.

11. An apparatus for compressing an original signal convertible into a plurality of binary signal vectors, each signal vector comprising a plurality of data elements of the original signal to be compressed, the apparatus comprising:

first data storage means for storing sets of binary code words;

first and second different sets of binary code words stored in the first data storage means;

means for hashing signal vectors of the plurality to identify uniquely for each hashed signal vector one subset of code words from one of the different sets of code words stored in the first storage means; and

second data storing means for storing as a compressed representation of the data of the original signal, a plurality of binary codes, each stored code being associated with one code word from the one subset of code words identified by the hashing means uniquely for each hashed signal vector.

12. The apparatus of claim 11 further comprising means for converting the original signal into the plurality of signal vectors.

13. A method of decompressing data compressed from an original signal and represented by a plurality of binary codes comprising the steps of:

identifying a first binary code word uniquely associated with a first code of the plurality, the first code word being longer than the first code;

identifying a second binary code word uniquely associated with a second code of the plurality, the second code word being longer than the second code and the second code being different in length from the first code; and

storing in a data storage device, a decompressed binary approximation of the data of the original signal using a combination of the first and second binary code words.

14. The method of claim 13 wherein the device includes a single, record storage element and the storing step further comprises storing on the single, record storage element the decompressed binary approximation of the data of the original signal.

15. A single record storage medium containing the decompressed binary approximation of the data of an original signal generated in accordance with the steps of the method of claim 14.

16. An apparatus for decompressing data compressed from an original signal represented by a plurality of binary codes of at least two different lengths comprising:

first data storage means for storing sets of binary code words;

at least first and second different sets of binary code words stored in the first data storage means, the first set having a first number of code words and the second set having a second number of code words

different from the first, each stored code word having a length longer than an associated binary code; and

means for identifying for each code one of the different sets of code words from the first data storage means containing the one code word associated with the code and identifying the one code word of the one identified set associated with the code and for converting the identified code words into decompressed data elements approximating data of the original signal.