Method for decoding signal vector-quantized by blocks

• US 5,903,679 A
• Filed: 04/14/1997
• Issued: 05/11/1999
• Est. Priority Date: 04/15/1996
• Status: Expired due to Term

• Current Assignee: Korea Telecommunication Authority
• Original Assignee: Korea Telecommunication Authority
• Inventors: Park, Seop Hyeong
• Primary Examiner(s): Boudreau, Leo H.
• Assistant Examiner(s): Tadayon, Bijan
• Application Number:
  US08/843,248
• Time in Patent Office: 757 Days
• Field of Search: 382/251, 382/252, 382/253, 382/270, 382/240, 382/199, 382/233, 382/250, 382/224, 382/115, 382/173, 382/232, 382/234, 382/235, 382/239, 382/244, 382/248, 382/260, 382/264 , 382/277, 382/278, 348/384, 348/607, 358/463, 358/430, 358/433, 358/539, 704/222, 704/223, 704/203, 704/204
• US Class Current: 382/253
• CPC Class Codes: G06T 9/008 Vector quantisation

Abstract

A method for producing a decoded signal approximate to an original signal upon decoding a quantized signal which is produced by vector-quantizing every block of an original signal or by vector-quantizing every block of a transform coefficient obtained after every block of the original signal is processed by a transform process such as a discrete cosine transform (DCT). The method includes the steps of determining a narrow vector quantization constraint set domain in which each vector or block divided from the decoded signal exists prior to the vector quantization at the transmitting site, and outputting a signal indicative of the determined narrow vector quantization constraint set domain, and performing a projection for returning a vector existing beyond the narrow vector quantization constraint set domain to its original domain, and repeating the projection for returning a vector existing beyond the narrow vector quantization constraint set domain to its original domain.

Claims

1. A method for producing a decoded image or audio signal by dividing, at a transmitting site, an image or audio signal into small blocks, vector-quantizing every block of the image or audio signal or every vector divided from the block of the image or audio signal, allocating a code to a representative value of the vector-quantized block or vector, and transmitting the code to a receiving site, while, at the receiving site, interpreting the received code, and reproducing a representative quantization vector based on the interpreted code, thereby obtaining a decoded image or audio signal, comprising the steps of:

transforming the received coded data into a codeword for each vector quantizer corresponding to the code;

transforming the codeword into a decoded signal having a re-constructed vector;

dividing the decoded signal into small blocks or vectors corresponding to those at the transmitting site, determining a narrow vector quantization constraint set domain in which each vector or block divided from the decoded signal exists prior to the vector quantization at the transmitting site, and outputting a signal indicative of the determined narrow vector quantization constraint set domain;

passing the decoded signal through a low pass filter, thereby reducing a blocking phenomenon occurring in the decoded signal;

applying the signal indicative of the narrow vector quantization constraint set domain, along with an output signal from the low pass filter, to a narrow vector quantization constraint set domain projector unit, thereby performing a projection for returning a vector existing beyond the narrow vector quantization constraint set domain to its original domain; and

applying a signal obtained after passing an output signal from the narrow vector quantization constraint set domain projector unit through the low pass filter, along with the signal indicative of the narrow vector quantization constraint set domain, to another narrow vector quantization constraint set domain projector unit, thereby repeating the projection for returning a vector existing beyond the narrow vector quantization constraint set domain to its original domain.

2. The method in accordance with claim 1, wherein the step of determining the narrow vector quantization constraint set domain comprises the step of determining a range, in which a vector value prior to the quantization exists, based on the representative quantization vector value prior to the decoding.

3. The method in accordance with claim 2, wherein the step of determining the narrow vector quantization constraint set domain comprises the step of re-constructing the domain in the form of a convex polytope so as to easily drive the value projected in each narrow vector quantization constraint set domain projector unit.

4. The method in accordance with claim 2, wherein the step of determining the narrow vector quantization constraint set domain comprises the step of varying the value projected in each narrow vector quantization constraint projector unit in accordance with each component of the vector.

5. The method in accordance with claim 1, wherein the narrow vector quantization constraint set domain includes a narrow vector quantization constraint set defined by the following equation:

space="preserve" listing-type="equation">C.sub.nvq ={x;

x.sub.n.sup.i ε

W.sub.n.sup.i }

where, "x_nⁱ " is an n-th vector, and "W_nⁱ " is a new domain obtained by reducing a convex polytope formed by reconstructing the narrow vector quantization constraint set domain.

6. The method in accordance with claim 1, wherein the narrow vector quantization constraint set domain includes a spherical narrow vector quantization constraint set domain defined as follows:

space="preserve" listing-type="equation">C.sub.SNVQ ={x;

x.sub.n ε

s.sub.n.sup.i, if O.sub.c (x.sub.n)=x.sub.n.sup.i ∀

n}

where, O_c is the vector quantizer, "x_n " is an n-th vector, and "S_nⁱ " is a new domain obtained by reducing a convex polytope formed by reconstructing the narrow vector quantization constraint set domain, the "S_nⁱ " being defined by the equation of "S_nⁱ ={x_n ;

∥

x_n -x_nⁱ ∥

≦

R_nⁱ }".

7. The method in accordance with claim 1, wherein the narrow vector quantization constraint set domain includes a cubic narrow vector quantization constraint set defined by the following equation:

space="preserve" listing-type="equation">C.sub.CNVQ ={x;

x.sub.n ε

c.sub.n.sup.i, if O.sub.c (x.sub.n.sup.i ∀

n}

where, O_c is the vector quantizer, "x_nⁱ " is an n-th vector, and "c_nⁱ " is a new domain obtained by reduced a convex polytope formed by reconstructing the narrow vector quantization constraint set domain.

8. A method for producing a decoded image or audio signal by dividing, at a transmitting site, an image or audio signal into small blocks, orthogonally transforming every block of the image or audio signal, vector-quantizing every orthogonally-transformed block or every vector divided from the orthogonally-transformed block, allocating a code to a representative value of the vector-quantized block or vector, and transmitting the code to a receiving site, while, at the receiving site, interpreting the received code, and reproducing a representative quantization vector based on the interpreted code, thereby obtaining a decoded image or audio signal, comprising the steps of:

transforming the received coded data into a codeword for each vector quantizer corresponding to the code;

transforming the codeword into a decoded signal having a re-constructed vector in a reverse quantizer unit;

dividing the decoded signal into small blocks or vectors corresponding to those at the transmitting site, determining a narrow vector quantization constraint set domain in which each vector or block divided from the decoded signal exists prior to the vector quantization at the transmitting site, and outputting a signal indicative of the determined narrow vector quantization constraint set domain;

re-constructing the vectors or blocks of the decoded signal into blocks having a size to be used for an orthogonal transform in a block constructing unit;

performing a reverse orthogonal transform for the re-constructed blocks in a reverse orthogonal transforming unit;

passing data obtained after the reverse orthogonal transform through a low pass filter;

orthogonally transforming an output signal from the low pass filter by blocks in an orthogonal transforming unit;

applying the signal indicative of the narrow vector quantization constraint set domain, along with an output signal from the low pass filter, to a narrow vector quantization constraint set domain projector unit and an output signal from the orthogonal transforming unit, thereby performing a projection for returning a vector existing beyond the narrow vector quantization constraint set domain to its original domain; and

applying a signal obtained after passing an output signal from the narrow vector quantization constraint set domain projector unit through the reverse orthogonal transforming unit, low pass filter unit and orthogonal transforming unit, along with the signal indicative of the narrow vector quantization constraint set domain, to another narrow vector quantization constraint set domain projector unit, thereby repeating the projection for returning a vector existing beyond the narrow vector quantization constraint set domain to its original domain.

9. The method in accordance with claim 8, wherein the step of determining the narrow vector quantization constraint set domain comprises the step of determining a range, in which a vector value prior to the quantization exists, based on the representative quantization vector value prior to the decoding.

10. The method in accordance with claim 8, wherein the step of determining the narrow vector quantization constraint set domain comprises the step of re-constructing the domain in the form of a convex polytope so as to easily drive the value projected in each narrow vector quantization constraint set domain projector unit.

11. The method in accordance with claim 8, wherein the step of determining the narrow vector quantization constraint set domain comprises the step of varying the value projected in each narrow vector quantization constraint projector unit in accordance with each component of the vector.