Efficient rate control for multi-resolution video encoding

Details Efficient rate control for multi-resolution video encoding Efficient rate control for multi-resolution video encoding

1998-02-26

2002-01-29

Kelley, Chris (Department: 2713)

Pulse or digital communications

Bandwidth reduction or expansion

Television or motion video signal

Reexamination Certificate

active

06343098

ABSTRACT:

FIELD OF THE INVENTION
The invention relates generally to video encoding which utilizes motion-compensated video compression techniques, and more particularly to rate control in multi-resolution video encoding applications.
BACKGROUND OF THE INVENTION
Motion video sequences typically contain a significant amount of intra-frame or “spatial” redundancy as well as inter-frame or “temporal” redundancy. Video compression techniques take advantage of this spatial and temporal redundancy to significantly reduce the amount of information bandwidth required to transmit, store and process video sequences. Existing standards for digital video compression include, for example, H.261, H.263, Motion-JPEG, MPEG-1 and MPEG-2. Transmission of compressed digital video can take place over many types of transmission facilities, and with many available bandwidths. For example, in a multipoint transmission application, two or more receivers of a compressed video bitstream may each have different available bandwidths with which to receive the video data. It is generally desirable in such an application to allow a receiver with a high bandwidth to receive higher resolution video than a receiver with a low bandwidth, rather than limiting all of the receivers to the low bandwidth. In these and other similar applications, a given video sequence is encoded at multiple resolutions.
The MPEG-2 standard implements multi-resolution video encoding through a process known as spatial scalability. This involves encoding a base layer of the video at a lower resolution and one or more enhancement layers at higher resolutions. The base layer is then transmitted to all receivers in a multipoint transmission application, and the enhancement layer or layers are transmitted only to the higher bandwidth receivers. However, MPEG-2 spatial scalability requires the higher bandwidth receiver to decode two or more layers, which increases the computational complexity of the decoding process. In addition, the bandwidth required for transmitting two or more layers is generally higher than that required for transmitting a single bitstream encoded at the higher resolution. Additional detail regarding these and other aspects of the MPEG-2 standard are described in greater detail in “Information Technology Generic Coding of Moving Pictures and Associated Audio Information: Video,” ISO/IEC DIS 13818-2, which is incorporated herein by reference.
FIG. 1
shows a conventional multi-resolution encoding system
10
. A video sequence in Common Intermediate Format (CIF) is supplied directly to a first standard video encoder
12
and also to a downsampler
14
. The first standard video encoder
12
encodes the CIF video sequence to generate a CIF bitstream. The downsampler
14
converts the CIF video sequence to a Quarter-CIF (QCIF) video sequence. A second standard video encoder
16
encodes the QCIF video sequence to generate a QCIF bitstream. The two encoders
12
,
16
operate substantially independently, and generally do not share rate control information.
FIG. 2
shows one of the standard video encoders
12
,
16
of
FIG. 1
in greater detail. The CIF or QCIF video sequence is applied via a signal combiner
20
to a discrete cosine transform (DCT) generator
22
which generates DCT coefficients for macroblocks of frames in the sequence. These coefficients are applied to a quantizer
24
, and the resulting quantized coefficients may be zig-zag scanned and run-amplitude coded before being applied to a variable-length coder (VLC)
26
. The output of the VLC
26
is an encoded bitstream. Rate control is provided by a rate control processor
28
. The DCT, quantization and variable-length coding operations of
FIG. 2
are designed to remove spatial redundancy within a given video frame in the sequence.
Temporal or inter-frame redundancy is removed in the encoder of
FIG. 2 through a
process of inter-frame motion estimation and predictive coding. For example, MPEG-2 video frames may be either intra-coded (I) frames, forward-only predictive (P) frames or bidirectionally-predictive (B) frames. An I frame is encoded using only the spatial compression techniques noted above, while a P frame is encoded using “predictive” macroblocks selected from a single reference frame. A given B frame is encoded using “bidirectionally-predictive” macroblocks generated by interpolating between a pair of predictive macroblocks selected from two reference frames, one preceding and the other following the B frame. In the encoder of
FIG. 2
, the output of the quantizer
24
is applied to an inverse quantizer
30
and then to an inverse DCT generator
32
. The output of the inverse DCT generator
32
is processed over one or more frames by a motion compensator
34
and motion estimator
36
. The motion compensator
34
generates motion vectors which are combined with a subsequent frame in signal combiner
20
so as to reduce inter-frame redundancy and facilitate encoding.
A conventional video encoder such as that shown in
FIG. 2
generally attempts to match the bitrate of the compressed video stream to a desired transmission bandwidth. The quantization parameter (QP) used in the quantizer
24
generally has a substantial effect on the resultant bitrate: a large QP performs coarse quantization, reducing the bitrate and the resulting video quality, while a small QP performs finer quantization, which leads to a higher bitrate and higher resulting image quality. The rate control processor
28
thus attempts to find a QP that is high enough to restrain the bitrate, but with the best possible resulting image quality. In general, it is desirable to maintain consistent image quality throughout a video sequence, rather than having the image quality vary widely from frame to frame. Both the MPEG-2 simulation model and the H.263 test model suggest rate control techniques for selecting the QP.
Approaches for implementing this type of rate control are described in greater detail in, for example, A. Puri and R. Aravind, “Motion-Compensated Video Coding with Adaptive Perceptual Quantization,” IEEE Transactions on Circuits and Systems for Video Technology, Vol. 1, No. 4, pp. 351-361, December 1991, and W. Ding and B. Liu, “Rate Control of MPEG Video Coding and Recording by Rate-Quantization Modeling,” IEEE Transactions on Circuits and Systems for Video Technology, Vol. 6, No. 1, pp. 12-20, February 1996, both of which are incorporated by reference herein. These approaches generally first select a target bitrate for each frame type (i.e., I frames, P frames and B frames), and the encoder attempts to assign the same number of bits to each frame of the same type. A frame-wide QP is then determined for each frame in an attempt to match the target bitrate for that frame. The approach described in the Puri and Aravind reference determines the frame-wide QP by using an activity measure, the frame variance. The approach described in the Ding and Liu reference generates a rate-quantization model. In either approach, the encoder may also vary the QP for individual macroblocks based on local activity measures.
A significant problem with these and other conventional rate control techniques is that they can be computation intensive, particularly for high resolution video sequences. For example, the approach in the Ding and Liu reference performs multi-pass encoding, that is, an entire frame is encoded more than one time using different QPs in order to find a QP that results in an actual bitrate closer to the target bitrate. This type of multi-pass encoding can be very computation intensive, and substantially reduces the efficiency of the encoding process.
SUMMARY OF THE INVENTION
The invention provides a multi-resolution video encoding system which improves the computational efficiency associated with encoding a video sequence in two or more different resolutions. An illustrative embodiment includes a first encoder for encoding the sequence at a first resolution, and a second encoder for encoding the sequence at a second resolution, where the second resolution is higher than the first reso

Affiliated with

Also associated with

Rating

Efficient rate control for multi-resolution video encoding does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Efficient rate control for multi-resolution video encoding, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Efficient rate control for multi-resolution video encoding will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-2859781