Quantization parameter adjustment based on sum of variance and estimated picture encoding cost

Patent metadata

Field	Value
Publication number	US-9565440-B2
Application number	US-201313926179-A
Country	US
Kind code	B2
Filing date	Jun 25, 2013
Priority date	Jun 25, 2013
Publication date	Feb 7, 2017
Grant date	Feb 7, 2017

Abstract

A video processing device includes a rate control module to determine more accurate initial quantization parameters at each scene switching point and to adjust the QP parameters in response to scene changes using a sum of variances metric and an estimated picture encoding cost metric from a coding complex estimation block. To determine a first quantization parameter set, a sum of variances metric and an estimated picture encoding cost metric for an initial set pictures of a video stream are used. A bit allocation module is to set a target bit allocation for infra-encoded pictures as substantially proportional to the sum of variances metric and substantially inversely proportional to the estimated picture encoding cost metric, and set a target bit allocation for forward predictive and bi-predictive pictures as substantially proportional to the estimated picture encoding cost metric and substantially inversely proportional to the sum of variances metric.

First claim

What is claimed is:

1. A method comprising:
determining, at a video processing device, a first quantization parameter set based on a sum of variances metric and an estimated picture encoding cost metric for each picture of an initial set of one or more pictures of a video stream;
encoding, at the video processing device, one or more pictures of the video stream using the first quantization parameter set;
setting a target bit allocation for intra-encoded pictures as substantially proportional to the sum of variances metric and substantially inversely proportional to the estimated picture encoding cost metric; and
setting a target bit allocation for forward predictive pictures and bi-predictive pictures as substantially proportional to the estimated picture encoding cost metric and substantially inversely proportional to the sum of variances metric.

2. The method of claim 1 , wherein:
the initial set of one or more pictures comprises one or more pictures immediately following a detected scene change; and
determining the first quantization parameter set comprises:
for each picture of the initial set, determining a sum of variances and an estimated picture encoding cost for the picture;
determining the sum of variances metric from the sums of variance of the pictures of the initial set; and
determining the estimated picture encoding cost metric from the estimated picture encoding costs of the pictures of the initial set.

3. The method of claim 2 , wherein:
determining the sum of variances metric as an average of the sums of variances; and
determining the estimated picture encoding cost metric as an average of the estimated picture encoding costs.

4. A method comprising:
determining, at a video processing device, a first quantization parameter set based on a sum of variances metric and an estimated picture encoding cost metric for each picture of an initial set of one or more pictures of a video stream;
encoding, at the video processing device, one or more pictures of the video stream using the first quantization parameter set;
wherein the first quantization parameter set comprises a first quantization parameter for encoding intra-encoded pictures, a second quantization parameter for encoding forward predictive pictures, and a third quantization parameter for encoding bi-predictive pictures;
wherein determining the first quantization parameter set comprises:
setting the first quantization parameter based on the sum of variances metric; and
setting the first and second quantization parameter based on the estimated picture encoding cost metric;

wherein setting the first quantization parameter based on the sum of variances metric comprises setting the first quantization parameter substantially proportional to a logarithmic function of the sum of variances metric and substantially inversely proportional to a logarithmic function of a target bit allocation for intra-encoded pictures;
wherein setting the second quantization parameter based on the estimated picture encoding cost metric comprises setting the second quantization parameter as substantially proportional to a logarithmic function of the estimated picture encoding cost metric and substantially inversely proportional to a logarithmic function of a target bit allocation for forward predictive pictures; and
wherein setting the third quantization parameter based on the estimated picture encoding cost metric comprises setting the third quantization parameter as substantially proportional to a logarithmic function of the estimated picture encoding cost metric and substantially inversely proportional to a logarithmic function of a target bit allocation for forward predictive pictures.

5. The method of claim 4 , further comprising:
adjusting the first quantization parameter based on a logarithmic function of a ratio of a sum of variances metric for a current picture to a sum of variances metric for a previous picture; and
adjusting the second quantization parameter based on a logarithmic function of a ratio of an estimated picture encoding cost metric of the current picture to an estimated picture encoding cost metric of the previous picture; and
adjusting the third quantization parameter based on a logarithmic function of a ratio of an estimated picture encoding cost metric of the current picture to an estimated picture encoding cost metric of the previous picture.

6. A video processing device comprising:
an interface to receive a video stream, the video stream comprising a sequence of pictures;
a rate control module to determine a first quantization parameter set based on a sum of variances metric and an estimated picture encoding cost metric for an initial set of one or more pictures of the video stream,
an encoder to encode one or more pictures of the video stream using the first quantization parameter set; and
a bit allocation module to:
set a target bit allocation for intra-encoded pictures as substantially proportional to the sum of variances metric and substantially inversely proportional to the estimated picture encoding cost metric; and
set a target bit allocation for forward predictive pictures and bi-predictive pictures as substantially proportional to the estimated picture encoding cost metric and substantially inversely proportional to the sum of variances metric.

7. The video processing device of claim 6 , wherein:
the initial set of one or more pictures comprises one or more pictures immediately following a detected scene change; and
the rate control module is to determine the first quantization parameter set by:
for each picture of the initial set, determining a sum of variances and an estimated picture encoding cost for the picture;
determining the sum of variances metric from the sums of variance of the pictures of the initial set; and
determining the estimated picture encoding cost metric from the estimated picture encoding costs of the pictures of the initial set.

8. The video processing device of claim 7 , wherein:
the sum of variances metric as an average of the sums of variances; and
the estimated picture encoding cost metric as an average of the estimated picture encoding costs.

Assignees

• Li Ying
• Zhao Xu Gang
• Vixs Systems Inc

Inventors

• Li Ying
• Zhao Xu Gang

Classifications

• H04N19/198

  including smoothing of a sequence of encoding parameters, e.g. by averaging, by choice of the maximum, minimum or median value · CPC title

• H04N19/142

  Detection of scene cut or scene change · CPC title

• H04N19/149

  by estimating the code amount by means of a model, e.g. mathematical model or statistical model · CPC title

• H04N19/124 Primary

  Quantisation · 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