Variable rate video encoder
The invention concerns a method of video encoding comprising: determining a first ratio (IP RATIO ) based on previously encoded picture frames of a video sequence comprising one or more intra-coded pictures (I) and one or more predicted pictures (P), the first ratio (IP RATIO ) being calculated based on the ratio between the size of one or more previously encoded intra-coded pictures and the size of one or more previously encoded predicted pictures; and determining a quantization parameter (QP) to be applied to a frame to be encoded based on the first ratio (IP RATIO ).
1. A method of video encoding comprising:
determining, using a rate control circuit, a first ratio (IPRATIO) based on previously encoded picture frames of a video sequence comprising one or more intra-coded pictures (I) and one or more predicted pictures (P), the first ratio (IPRATIO) being calculated based on the ratio between the size of one or more previously encoded intra-coded pictures and the size of one or more previously encoded predicted pictures;
determining, using the rate control circuit, the size (SSW) of a sliding window of N previously encoded picture frames of the video sequence, the sliding window comprising one or more of the intra-coded pictures and one or more of the predicted pictures, where N is an integer equal to three or more;
determining, using the rate control circuit, a target size (TSW) of the sliding window based on a target size of a group (GOP) of successive pictures comprising a number of pictures from one intra-coded picture up to but not including a subsequent intra-coded picture, the target size being determined based on the following equation:
T
SW
=
(
N
I
_
SW
×
IP
RATIO
+
N
P
_
SW
)
×
GOP
TARGET
_
SIZE
IP
RATIO
+
N
P
_
GOP
where N I _ SW is the number of intra-coded pictures in the sliding window, N P _ SW is the number of predicted pictures in the sliding window, GOP TARGET _ SIZE is the target size of the group of successive pictures, IP RATIO is the first ratio and N P _ GOP is the number of predicted pictures in the group of successive pictures;
determining, using the rate control circuit, a quantization parameter (QP) to be applied to a frame to be encoded based on the first ratio (IPRATIO) and a comparison of the size of the sliding window with the target size of the sliding window; and
encoding, using an encoder, the frame to be encoded using the determined quantization parameter.
2. The method of claim 1 , wherein the first ratio (IP RATIO ) is determined based on a previous value of the first ratio (PREVIOUS_IP RATIO ) and on a latest ratio between the encoded size of the latest of the intra-coded pictures to be encoded and the latest of the predicted pictures to be encoded.
3. The method of claim 2 , wherein the first ratio is determined based on the following equation:
IP
RATIO
=
PREVIOUS_IP
RATIO
W
+
LATEST_IP
RATIO
COUNT_IP
RATIOS
where PREVIOUS_IP RATIO is the previous value of the first ratio, W is a weight applied to the previous value of the first ratio, LATEST_IP RATIO is said latest ratio and COUNT_IP RATIOS is a value representing the number of ratios determined based on the following equation:
COUNT_IP
RATIOS
=
PREVIOUS_COUNT
_IP
RATIOS
W
+
1
where PREVIOUS_COUNT_IP RATIOS is the previous value of COUNT_IP RATIOS .
4. The method of claim 1 , wherein:
the sliding window further comprises one or more bi-predicted pictures (B); and
determining the target size (T SW ) of the sliding window is further based on a second ratio (PB RATIO ) representing the ratio between the size of one or more of said predicted pictures and the size of one or more of said bi-predicted pictures.
5. The method of claim 1 , wherein the target size of the group of successive pictures is determined based on the following equation:
GOP
TARGET
_
SIZE
=
BITRATE
TARGET
×
GOP_LENGTH
FRAME_RATE
where BITRATE TARGET is a target bitrate of the encoded video stream, GOP_LENGTH is the number of pictures in the group of pictures, and FRAME_RATE is the frame rate of the video sequence.
6. The method of claim 1 , wherein determining a quantization parameter (QP) to be applied to a frame to be encoded comprises determining a correction factor to be applied to a previous quantization parameter by dividing the size of the sliding window by the target size of the sliding window.
7. The method of claim 6 , wherein the quantization parameter has a logarithmic relation with the bitrate, and the quantization parameter is determined by applying the correction factor to a linearized value (QS) determined by linearizing the previous quantization parameter.
8. The method of claim 6 , wherein the quantization parameter (NEWQP) is determined based on the following equation:
NEWQP= f QP ( f QS (OLDQP)×CORRECTION_FACTOR)
where OLDQP is the previous quantization parameter, CORRECTION_FACTOR is the correction factor, f QS is a linearization function for linearizing the previous quantization parameter and f QP is the inverse of the linearization function.
9. A video system comprising:
a rate control circuit configured to:
determine the size (S SW ) of a sliding window of N previously encoded picture frames of the video sequence, the sliding window comprising one or more of the intra-coded pictures and one or more of the predicted pictures, where N is an integer equal to three or more;
determine a target size (T SW ) of the sliding window based on a target size of a group (GOP) of successive pictures comprising a number of pictures from one intra-coded picture up to but not including a subsequent intra-coded picture, the target size being determined based on the following equation:
T
SW
=
(
N
I
_
SW
×
IP
RATIO
+
N
P
_
SW
)
×
GOP
TARGET
_
SIZE
IP
RATIO
+
N
P
_
GOP
where N I _ SW is the number of intra-coded pictures in the sliding window, N P _ SW is the number of predicted pictures in the sliding window, GOP TARGET _ SIZE is the target size of the group of successive pictures, IP RATIO is the first ratio and N P _ GOP is the number of predicted pictures in the group of successive pictures;
determine a first ratio (IP RATIO ) based on previously encoded picture frames of a video sequence comprising one or more intra-coded pictures (I) and one or more predicted pictures (P), the first ratio (IP RATIO ) being calculated as the ratio between the size of one or more previously encoded intra-coded pictures and the size of one or more previously encoded predicted pictures; and
determine a quantization parameter (QP) to be applied to a frame to be encoded based on the first ratio (IP RATIO ) and a comparison of the size of the sliding window with the target size of the sliding window; and
an encoder configured to encode the frame to be encoded using the determined quantization parameter.
10. The video system of claim 9 , wherein:
the sliding window further comprises one or more bi-predicted pictures (B); and
the rate control circuit is further configured to determine the target size (T SW ) of the sliding window based on a second ratio (PB RATIO ) representing the ratio between the size of one or more of said predicted pictures and the size of one or more of said bi-predicted pictures.
11. The video system of claim 9 , wherein the rate control circuit is further configured to determine the first ratio (IP RATIO ) based on a previous value of the first ratio (PREVIOUS_IP RATIO ) and on a latest ratio between the encoded size of the latest of the intra-coded pictures to be encoded and the latest of the predicted pictures to be encoded.