Quantization matrix compression in video coding

Abstract

A method of quantization matrix compression in a video encoder is provided that includes preprocessing a quantization matrix by performing at least one selected from down-sampling the quantization matrix and imposing 135 degree symmetry on the quantization matrix, performing zigzag scanning on the pre-processed quantization matrix to generate a one dimensional (1D) sequence, predicting the 1D sequence to generate a residual 1D sequence, and coding the residual 1D sequence using kth order exp-Golomb coding to generate a compressed quantization matrix, wherein k≥0.

What is claimed is: 
     
         1 . A method of decoding an encoded video bit stream, comprising:
 decoding the encoded video bit stream into an entropy decoded bit stream including decoded bits, the decoded bits comprising quantized transform coefficients;   extracting a first flag from the decoded bits, the first flag indicating a non-default quantization matrix;   extracting a second flag from the decoded bits, the second flag indicating a relationship between a first and second index position where the non-default quantization matrix will be implemented; and   performing an inverse quantization operation on the quantized transform coefficients using the non-default quantization matrix at either the first or second index position.   
     
     
         2 . The method of  claim 1 , wherein performing an inverse quantization operation on the quantized transform coefficients using the non-default quantization matrix at either the first or second index position results in transform coefficients; and
 further comprising performing an inverse transform operation on the transform coefficients to form reconstructed non-over-lapping blocks.   
     
     
         3 . The method of  claim 2  further comprising forming a frame using the reconstructed non-over-lapping blocks and displaying the frame on a display. 
     
     
         4 . The method of  claim 1  further comprising recovering the non-default quantization matrix in a first portion of the entropy decoded bit stream. 
     
     
         5 . The method of  claim 4  wherein the first portion of the entropy decoded bit stream is for sequence parameter sets (SPS). 
     
     
         6 . The method of  claim 4  wherein the first portion of the entropy decoded bit stream is for picture parameter sets (PPS). 
     
     
         7 . The method of  claim 1  wherein the relationship is a difference. 
     
     
         8 . The method of  claim 1  wherein the non-default quantization matrix is an 4×4 matrix, an 8×8 matrix or a 16×16 matrix. 
     
     
         9 . A system for decoding an encoded video bit stream, comprising:
 an entropy decoding component configured to entropy the encoded video bit stream into decoded bits, the decoded bits comprising quantized transform coefficients; and   an inverse quantization component configured to:
 extract a first flag from the decoded bits, the first flag indicating a non-default quantization matrix; 
 extract a second flag from the decoded bits, the second flag indicating a relationship between a first and second index position where the non-default quantization matrix will be implemented; 
 perform an inverse quantization operation on the quantized transform coefficients using the non-default quantization matrix at either the first or second index position. 
   
     
     
         10 . The system of  claim 9 , wherein performing an inverse quantization operation on the quantized transform coefficients using the non-default quantization matrix at either the first or second index position results in transform coefficients; and
 the inverse quantization component configured to perform an inverse transform operation on the transform coefficients to form reconstructed non-over-lapping blocks.   
     
     
         11 . The system of  claim 9  wherein the entropy decoding component is further configured to recover the non-default quantization matrix in a first portion of the entropy decoded video bit stream. 
     
     
         12 . The system of  claim 11  wherein the first portion of the decoded bit stream is for sequence parameter sets (SPS). 
     
     
         13 . The system of  claim 11  wherein the first portion of the entropy decoded bit stream is for picture parameter sets (PPS). 
     
     
         14 . The system of  claim 9  wherein the relationship is a difference. 
     
     
         15 . The system of  claim 9  wherein the non-default quantization matrix is an 4×4 matrix, an 8×8 matrix, or a 16×16 matrix. 
     
     
         16 . A method comprising:
 decoding, by a decoder, an encoded video bit stream into quantized transform coefficients;   determining, from a first indicator in the encoded video bit stream, the quantized transform coefficients are associated with a non-default quantization matrix;   determining, from a second indicator in the encoded video bit stream, a reference quantization matrix based on a difference between a first scaling list index of the non-default quantization matrix and a second scaling list index of the reference quantization matrix; and   performing an inverse quantization operation on the quantized transform coefficients based on the non-default quantization matrix and the reference quantization matrix.   
     
     
         17 . The method of  claim 16 , wherein:
 the non-default quantization matrix is included in one of a sequence parameter set (SPS) and a picture parameter set (PPS).   
     
     
         18 . The method of  claim 17 , wherein:
 a 1-bit flag is included in the PPS to indicate to the decoder a source of the reference quantization matrix; and   the source is one of the SPS or the PPS.   
     
     
         19 . The method of  claim 18 , wherein:
 the source is the SPS; and   the 1-bit flag is encoded with a signed exp-Golomb code.   
     
     
         20 . The method of  claim 18 , wherein:
 the source is the PPS; and   the 1-bit flag is encoded with an unsigned exp-Golomb code.