Single-pass constant- and variable-bit-rate MPEG-2 video compression

IBM Journal of Research and Development, Jul 1999 by Mohsenian, N, Rajagopalan, R, Gonzales, C A

Most real-time MPEG-2 encoders are designed to perform in a constant-bit-rate (CBR) mode, in which buffer constraints are imposed to circumvent large deviations from a desired rate at any instant in time. Although such streams are generally good-quality sequences, certain types of operations or environments call for a more efficient real-time CBR encoder. The first part of the paper describes how a better-quality CBR video stream can be produced by estimating the relative complexity of a picture in comparison with the average complexity of the partially encoded stream and using it to adjust the compression parameters in a single-pass mode of operation. Our CBR encoder is particularly attractive for digital broadcast and editing environments, in which representations of higher-fidelity video objects in both display and freeze modes are constantly pursued. The second part of the paper describes the real-time generation of video streams with a variable-bit-rate (VBR) encoder. This mode of operation is highly desirable for home entertainment and recreational events. We propose a robust single-pass VBR video encoder algorithm which is capable of learning and adapting itself to the complexity of image segments and thereafter creating streams which have constant visual picture quality. The new VBR scheme displays a better performance than the CBR encoder, particularly when special effects such as scene transitions, fades, or luminance changes are to be compressed. Both CBR and VBR encoders are fully compliant with the MPEG-2 standard and are easily implementable with IBM encoder architecture. Compression results for the new single-pass encoding algorithms and comparisons with previous CBR schemes are provided. The result suggests the suitability of our VBR approach for record/playback in storage media such as digital video disc (DVD) players, disk-based camcorders, and digital videocassette recorders (DVCRs). It further reflects the importance of our single-pass CBR scheme for providers of broadcast services, for which it allows more video programs to be allocated to a selected communication link, and for in-studio applications, for which it greatly facilitates visual analysis of captured streams.

Introduction

The standardization of MPEG-2 [1] has greatly facilitated the transmission, representation, storage, and manipulation of digital video in various environments such as broadcast television, wireless communications, consumer electronics, and multimedia computers. Further, applications ranging from desktop publishing on personal computers (PCs) to home authoring with digital video disc (DVD) players demonstrate the major role played by the MPEG-2 standard in promoting new storage media for consumer video and interactive multimedia. This has enabled the home user to download video streams from a satellite system or an Internet site in order to create DVD video programs or multimedia presentations using a recordable medium. Although the syntax and specification of MPEG-2 bitstreams and multimedia programs (as in DVD titles) are well defined by the international standards, the actual encoding parameters and functions, which should lead to a fully compliant bitstream, have been and are the subject of many research efforts. The main challenge is how to achieve a close-to-optimum video quality in a compressed stream while reducing the amount of information in the source. This is because the statistical nature of any video source is either not known a priori or will change over time, and a true estimation of source distribution can be anywhere from computationally extensive, as in non-real-time multipass encoding, to almost impossible for real-time encoding.

The nonstationary nature of images makes them inherently variable- compression optimality for MPEG-2 coder-decoders (codecs) is achieved by carefully selecting a set of spatial or temporal image analyzers, quantizers, and variable-length entropy coders, of which some are frozen by the standard and some are to be defined by the designer. The results are variable' streams which require a sophisticated buffering scheme to smooth out the variability of the signal before transmission over a fixed bandwidth is carried out. The receiver will have a similar buffering policy to convert the fixed-channel rate to variable streams prior to the decoding and display of each picture.

Since design and development of an MPEG-2 video encoder can become cumbersome as a result of formulating several mathematically or perceptually derived parameters, typical approaches [21 enforce a constant bit rate (CBR) for a group of pictures (GOP) regardless of the complexity of the video interval. This scheme assumes equal weighting of bit distribution among GOPs and reduces the degree of freedom of the encoding task. In short, the problem is reduced to minimizing (maximizing) the GOP distortion (quality) subject to a constant target rate. By CBR we mean that the sustainable rate of the encoded video stream per GOP is close to a constant target rate, but the instantaneous rate changes per picture depending on picture type or the quantization sealer. Another advantage of a CBR stream is that the transmitted signal may be terminated at any time and the user is assured of maintaining a rate close to the target rate. All CBR MPEG-2 encoders enforce different quantizing sealers for each picture type to achieve good-quality streams within a GOP. This method of compression works adequately when the complexity of the source varies slowly over time and therefore the encoding algorithm has time to adjust itself. However, if the statistical features of the source change rapidly over time, a constant-bit-rate operation may result in good picture quality for a short time window (e.g., a few frames or a GOP) and discontinuous quality when the whole video is perceived.