Abstract

Many video services use pre-encoded video for the distribution of video programs to end users. The transmission of compressed video over channels with different capacities may require a reduction in bit rate if the transmission media has a lower capacity than the capacity required by the video bitstream, or when the communication network is congested. The process of converting a compressed video format into another compressed format is known as transcoding. This thesis addresses the specific transcoding problem of bitrate reduction of a previously compressed MPEG video. Fully decoding a compressed video then re-encoding it at a lower bit rate, as a second generation video, has two disadvantages. First, it is not an efficient solution in terms of implementation complexity, delay and cost. Second, errors are introduced in the repeated compression/decompression of MPEG video, known as multigeneration. In this research, five mechanisms contributing to the continued degradation in multigeneration of MPEG video are identified: Pixel Domain Quantization (PDQ), Pixel Domain Clipping (PDC), Compression Control Parameters Variation (CCPV), Motion Vector Re-estimation (MVR) and Error Propagation due to Motion Compensation (EPMC). The degradation caused by each mechanism is illustrated and quantified by experiments. Next, the research addresses transcoding of MPEG compressed video. Two methods to reduce the requantization errors in transcoding are proposed. The first method assumes Laplacian distributions for the original DCT coefficients. A Laplacian parameter for each coefficient is estimated at the transcoder from the quantized input DCT coefficients. These parameters are used in transcoding to improve the quality of the transcoded video. The second method, selective requantization, is based on avoiding critical ratios of the two cascaded quantizations (encoding versus transcoding) that either lead to larger transcoding errors or require a higher bit budget. The experimental results show that both methods improve the quality of the transcoded video. Moreover, the thesis addresses the problem of multi-program video transmission over heterogeneous networks and provides a joint transcoder for transcoding multiple MPEG video bitstreams simultaneously. It is shown that joint transcoding provides better picture quality than independent transcoding of each sequence at a constant bitrate. Furthermore, joint transcoding minimizes the variation in picture quality between the sequences, as well as within each sequence. Consequently, joint transcoding results in a better utilization of the channel capacity.