16-Bit TransformsMotion CompensationEE 5359 PROPOSALH.264 to VC-1 TRANSCODING Vidhya VijayakumarStudent I.D.: 1000-622152Date: September 24, 20091H.264 to VC-1 TRANSCODEROBJECTIVE:The objective of the thesis is to implement a H.264 bitstream to VC-1 transcoder for progressive compression. MOTIVATION:The high definition video adoption has been growing rapidly for the last fiveyears. The high definition DVD format Blue ray has mandated MPEG-2[3], H.264 [2]and VC-1 [1] as video compression formats. The coexistence of these different videocoding standards creates a need for transcoding. As more and more end products usethe above standards, transcoding from one format to another adds value to theproduct’s capability. While there has been recent work on MPEG-2 to H.264transcoding [3], VC-1 to H.264 transcoding [4], the published work on H.264 to VC-1transcoding is nearly non-existent. This has created the motivation to develop atranscoder that can efficiently transcode a H.264 bitstream to a VC-1 bitstream. DETAILS:Video transcoding is the operation of converting video from one format toanother [5]. A format is defined by characteristics such as bit-rate, spatial resolutionetc. One of the earliest applications of transcoding is to adapt the bit-rate of acompressed stream to the channel bandwidth for universal multimedia access in allkinds of channels like wireless networks, Internet, dial-up networks etc. Changes inthe characteristics of an encoded stream like bit-rate, spatial resolution, quality etc canalso be achieved by scalable video coding [5].However, in cases where the availablenetwork bandwidth is insufficient or if it fluctuates with time, it may be difficult to setthe base layer bit-rate. In addition, scalable video coding demands additionalcomplexities at both the encoder and the decoder.The basic architecture for converting an H.264 bitstream into a VC-1elementary stream arises from complete decoding of the H.264 stream and then re-encoding into a VC-1 stream. However, this involves significant computationalcomplexity [6]. Hence there also is a need to transcode at low complexity. Transcoding can in general be implemented in the spatial domain or in thetransform domain or in a combination of the two domains. The common transcodingarchitectures [5] are: Open loop transform domain transcoding2Fig. 1 Open loop transform domain transcoder architecture [5]Open loop transcoders are computationally efficient (Fig 1). They operate in the DCTdomain. However they are subject to drift error. Drift error occurs due to rounding,quantization loss and clipping functions. Cascaded Pixel Domain Architecture (CPDT)Fig. 2 Cascaded pixel domain transcoder architecture [5]This is the most basic transcoding architecture (Fig 2). The motion vectors from theincoming bit stream are extracted and reused. Thus the complexity of the motionestimation block is eliminated which accounts for 60% of the encoder computation.As compared to the previous architecture, CPDT is drift free. Hence, even though it isslightly more complex, it is suited for heterogeneous transcoding between differentstandards where the basic parameters like mode decisions, motion vectors etc are tobe re-derived.Simplified DCT Domain transcoders (SDDT)Fig. 3 Simplified transform domain transcoder architecture [5]3This transcoder is based on the assumption that DCT, IDCT and motion compensationare linear processes (Fig 3). This architecture requires that motion compensation beperformed in the DCT domain, which is a major computationally intensive operation[3]. For instance, as shown in the figure 4, the goal is trying to compute the DCTcoefficients of the target block B from the four overlapping blocks B1, B2, B3 andB4.Fig. 4 Transform domain motion compensation illustration [5]Also, clipping functions and rounding operations performed for interpolation in fractional pixel motion compensation lead to a drift in the transcoded video. Cascaded DCT Domain transcoders (CDDT)Fig. 5 Cascaded transform domain transcoder architecture [5]This is used for spatial/temporal resolution downscaling and other coding parameterchanges (Fig 5). As compared with SDDT, greater flexibility is achieved byintroducing another transform domain motion compensation block; however it is farmore computationally intensive and requires more memory [3]. It is often applied todownscaling applications where the encoder end memory will not cost much due todownscaled resolution.4Choice of basic transcoder architecture: DCT domain transcoders have the main drawback that motion compensationin transform domain is very computationally intensive. DCT domain transcoders arealso, less flexible as compared to pixel domain transcoders, for instance, the SDDTarchitecture can only be used for bit rate reduction transcoding. It assumes that thespatial and temporal resolutions stay the same and that the output video uses the sameframe types, mode decisions and motion vectors as the input video. For H.264 to VC-1 transcoding, it is required to implement several changes inorder to accommodate the mismatches between the two standards. For instance, formotion estimation and compensation, H.264 supports 16x16, 16x8, 8x16, 8x8, 8x4,4x8, 4x4 macroblock partitions (Fig 6), but VC-1 supports 16x16 and 8x8 only (Fig7). The transform size and type (8x8 and 4x4 in H.264 and 8x8, 4x8, 8x4 and 4x4 inVC-1) are different and make transform domain transcoding prohibitively complex.Hence, the use of DCT domain transcoders is not very ideal.Fig.6 Segmentations of the macroblock for motion compensation in H.264Top: segmentation of macroblocks, bottom: segmentation of 8x8 partitions. [2] Fig.7 Segmentations of the macroblock for motion compensation in VC-1 [2]From Fig. 8, it can be inferred that, the cascaded pixel domain architectureoutperforms the DCT domain transcoders. Also for larger GOP sizes, the drift in DCTdomain transcoders becomes more significant.5Fig.8 PSNR vs Bit-rate graph for the Foreman sequence transcoded with a GOP size 15, usingdifferent transcoding architectures as described in Figs. 1, 2, 3 and 5. [5]Hence, heterogeneous transcoding in the pixel domain is preferred for standards transcoding.Standards transcoding: When transcoding between two different standards, the main
View Full Document
Unlocking...