Rate-Distortion Optimization in Dynamic Mesh Compression

Rate-distortion-optimized predictive compression of dynamic 3D mesh sequences

H.264/AVC is the latest international video coding standard developed by ITU-T Video Coding Expert Group and the ISO/IEC Moving Picture Expert Group, which provides gains in compression efficiency of about 40% compared to previous standards (ISO/IEC 14496-10, 2004, Weigand et al., 2003). New and advanced techniques are introduced in this new standard, such as intra prediction for I-frame encoding, multi-frame inter prediction, small block-size transform coding, context-adaptive binary arithmetic coding (CABAC), deblocking filtering, etc. These advanced techniques offer approximately 40% bit rate saving for comparable perceptual quality relative to the performance of prior standards (Weigand ct al., 2003). H.264 intra prediction offers nine prediction modes for 4x4 luma blocks, nine prediction modes for 8x8 luma blocks and four prediction modes for 16 x 16 luma blocks. However, the rate-distortion (RD) performance of the intra frame coding is still lower than that of inter frame coding. Hence intra frame coding usually requires much larger bits than inter frame coding which results in buffer control difficulties and/or dropping of several frames after the intra frames in real-time video. Thus the development of an efficient intra coding technique is an important task for overall bit rate reduction and efficient streaming. H.264/AVC uses rate-distortion optimization (RDO) technique to get the best coding mode out of nine prediction modes in terms of maximizing coding quality and minimizing bit rates. This means that the encoder has to code the video by exhaustively trying all of the nine mode combinations. The best mode is the one having the minimum rate-distortion (RD) cost. In order to compute RD cost for each mode, the same operation of forward and inverse transform/quantization and entropy coding is repetitively performed. All of these processing explains the high complexity of RD cost calculation. Therefore, computational complexity of encoder is increased drastically. Using nine prediction modes in intra 4x4 and 8x8 block unit for a 16x16 macroblock (MB) can reduce spatial redundancies, but it may needs a lot of overhead bits to represent the prediction mode of each 4x4 and 8x8 block. Fast intra mode decision algorithms were proposed to reduce the number of modes that needed calculation according to some criteria (Sarwer et al.,2008, Tsai et al., 2008, Kim, 2008, Pan et al., 2005, Yang et al., 2004). An intra mode bits skip (IBS) method based on adaptive singlemultiple prediction is proposed in order to reduce not only the overhead mode bits but also computational cost of the encoder (Kim et al., 2010). If the neighbouring pixels of upper and left blocks are similar, only DC prediction is used and it does not need prediction mode bits or else nine prediction modes are computed. But the IBS method suffers with some drawbacks a) the reference pixels in up-right block are not considered for similarity

Rate distortion (RD) optimization, which is used to decide the best coding mode of a macroblock (MB) in H.264/AVC video coding standard, is a very efficient technique to improve coding performance. On the other hand, this RD optimization process also incurs a considerable complexity increase of the encoder. To reduce computational cost of RD optimization, fast bit estimation method can be adopted to avoid entropy coding computation during the mode decision process of H.264/AVC encoding [1] [2]. However, previous works [1][2] are all for CAVLC (Context Based Adaptive Variable Length Coding) based RD optimization process and are not accurate for CABAC (Context Based Adaptive Binary Arithmetic Coding) based RD optimization process. As known, compared with CAVLC, CABAC entropy coding provides higher coding efficiency at average bit rate saving of 9%-14% [3]. Hence, it is desirable to find a fast bit estimation method that is suitable for RD optimization with CABAC entropy coding. In this paper, we present such kind of fast bit estimation method. The experiments show that the proposed method can reduce around 30% of the RD optimization time averagely while maintaining the similar coding performance comparing to that of original H.264/AVC with RD optimization enabled.

Rate distortion (RD) optimization for H.264 interframe coding with complete baseline decoding compatibility is investigated on a frame basis. Using soft decision quantization (SDQ) rather than the standard hard decision quantization, we first establish a general framework in which motion estimation, quantization, and entropy coding (in H.264) for the current frame can be jointly designed to minimize a true RD cost given previously coded reference frames. We then propose three RD optimization algorithms--a graph-based algorithm for near optimal SDQ in H.264 baseline encoding given motion estimation and quantization step sizes, an algorithm for near optimal residual coding in H.264 baseline encoding given motion estimation, and an iterative overall algorithm to optimize H.264 baseline encoding for each individual frame given previously coded reference frames-with them embedded in the indicated order. The graph-based algorithm for near optimal SDQ is the core; given motion estimation and quantization step sizes, it is guaranteed to perform optimal SDQ if the weak adjacent block dependency utilized in the context adaptive variable length coding of H.264 is ignored for optimization. The proposed algorithms have been implemented based on the reference encoder JM82 of H.264 with complete compatibility to the baseline profile. Experiments show that for a set of typical video testing sequences, the graph-based algorithm for near optimal SDQ, the algorithm for near optimal residual coding, and the overall algorithm achieve on average, 6%, 8%, and 12%, respectively, rate reduction at the same PSNR (ranging from 30 to 38 dB) when compared with the RD optimization method implemented in the H.264 reference software.

To ensure the fidelity of virtual views, rate-distortion optimization (RDO) criterion for the 3D extension of the High Efficiency Video Coding (3D-HEVC) is well designed, in which the synthesized view distortion (SVD) is introduced to derive the rate-distortion (RD) cost. To obtain accurate SVDs, the rendering operation is employed which demands a fairly high computational complexity. To address this problem, a fast RDO method for depth maps is proposed, which checks the RD cost during its calculation process. Specifically, given a coding mode, the RD cost is composed of several cumulative items. If the accumulated RD cost is equal to or exceeds the minimum RD cost of previously coded modes, it will not be necessary to continue the RD cost calculation for the mode. To reduce the encoding complexity, existing methods usually aim at reducing the number of tested modes or block partitions. To the best of our knowledge, it is the first time that the latent redundant complexity in the RD cost calculation is investigated and removed. Experimental results demonstrate that, compared with the 3D-HEVC reference software, the proposed method can save 28.1% of depth coding time with a small coding gain (0.04% BD-rate saving). An additional test is designed to evaluate four typical fast coding methods with/without the proposed method. Extensive results verify that the proposed method can be seamlessly combined with the state-of-the-art methods.

In High Efficiency Video Coding (HEVC), the computational complexity has been increased so that an Early SKIP mode decision method is proposed by using coded block flag of an inter prediction unit (PU) to speed up mode decision with BD-bitrate increase. In this paper, we propose a fast SKIP mode decision algorithm to speed up PU mode decision for HEVC in a rate-distortion (RD) optimization sense. It is based on an adaptive linear predictor for PU mode type and available neighboring SKIP mode units on the quad-tree partition structure in HEVC for adaptive RD optimization. Moreover, the proposed method consists of both local (neighboring PUs) and global (previous PUs in current frame) level schemes on each coding unit (CU) level. In this way, it is possible to exploit high correlations between the RD cost of SKIP mode and spatial characteristics of video sequences by using the distribution of SKIP mode and a linear RD cost predictor. Experimental results show that the encoding computational complexity can be reduced by an average of 36 % with threshold RD cost, when compared to that of the HM 12 reference software.

The H.264/AVC video coding standard can achieve better compression performance than previous video coding standards, such as MPEG-2, MPEG-4, and H.263, because H.264 encoder employs a rate-distortion (RD) optimization strategy to select the best coding mode. With this strategy, the RD costs for all possible modes are computed and the one with the minimum RD cost is considered the best mode. However, this coding step results in very high computational complexity. Therefore, in this thesis, we propose a fast mode decision algorithm based on mode statistical information and the motion cost of P16×16 mode. The mode statistics information provides the possibility that the mode for a macroblock (MB) is SKIP mode, given that the best modes for the MB’s in some neighborhood of that MB are all SKIP modes. For each MB, the proposed algorithm checks the mode statistical information first in order to terminate the RD optimization calculation as early as possible by using SKIP mode directly to encode this MB. Otherwise, the motion cost of P16×16 mode is obtained to determine which mode category is more likely to contain the true best mode, and the required number of modes considered in the best mode prediction process can be reduced significantly. Experimental results demonstrate that, on average, the proposed approach can save the encoding time by 78.03% and encoding bit rate by 2.06%, while incurring only 0.64 dB drop in PSNR.

The H.264/AVC video coding standard can achieve higher compression performance than previous video coding standards, such as MPEG-2, MPEG-4, and H.263. In order to improve the coding performance, the H.264/AVC encoder employs various modes and the rate-distortion (RD) optimization method for selection of an optimum mode. Since the encoder computes RD costs of all possible coding modes to decide the optimum mode, it induces high computational complexity. In this paper, we introduce a fast mode decision method for inter picture macroblocks to reduce the computational complexity. The proposed method significantly reduces the number of candidate modes for the RD optimization process by detecting spatially and temporally homogeneous regions and analyzing motion costs for inter modes and intra prediction costs for intra modes. Experimental results demonstrate that the proposed method drastically reduces the encoding time and the number of the RD cost calculation process while maintaining the high coding efficiency.

A novel approach to hardware implementation of H.264 block size decision is proposed, which is based on rate-distortion (RD) optimization. Utilization of RD cost for block size decision can improve up to 2.0 dB of PSNR in compared with conventional SAD/SATD based approaches. However, calculation of RD cost for a block incurs considerable computational costs since distortion can be determined only after completing the whole encoding processes of the block. Thus the proposed approach simplifies VLC process and our hardware employs 7 stage pipeline architecture for the cost calculation. As a result, the proposed architecture, which can be implemented by 20k gates, achieves real-time processing of SD (720times480) frames at a rate of 30 fps in 23.7 MHz operation

Advanced motion vector prediction (AMVP) is a new technique adopted in the latest high efficiency video coding (HEVC) standard. AMVP block predicts an initial motion vector of the current block from a given set of candidates by means of rate distortion (RD) optimization process. Due to the large number of different-sized blocks, simplification of RD optimization process in AMVP block is highly appreciated. Therefore, we present a new RD optimization technique for AMVP block in HEVC encoder. The proposed RD calculation approach finds the best AMVP candidate by processing less number of feature pixels per every block. Experimental results show notable speedup in terms of AMVP processing time with tolerable quality degradation (PSNR) and bitrate requirement. The proposed RD calculation technique reduces the RD computational complexity of the AMVP block by 87.5% as maximum (i.e. 1.7% of the whole encoder complexity). This improvement is accompanied with a modest average PSNR loss of 0.10 dB and an increase by 2.4% in terms of bitrate. On the other hand, we present an FPGA-based architecture for AMVP unit in HEVC encoder. The proposed architecture was prototyped, simulated and synthesized on Xilinx Virtex-7 XC7VX550T FPGA. At 188 MHz clock frequency, the proposed architecture processes 8 K (7680 \(\times \) 4320) YCrCb resolution at 60 fps while utilizing less than 1% of the FPGA resources.

High Efficiency Video Coding (HEVC) intra coding achieves significant improvements in coding efficiency compared with H.264/AVC intra coding by adopting 35 spatial intra prediction modes and a quadtree-based block partitioning structure. However, the encoding complexity is extremely high from performing rate-distortion optimization (RDO) on each mode and depth level. This paper proposes fast intra mode decision algorithms for real-time HEVC encoding. For a fast intra prediction mode decision, we estimate the lower bound on the rate-distortion (RD) cost of a prediction unit with non–most-probable modes (non-MPMs). If the RD cost of the MPM is lower than the lower bound, the MPM is selected as the best mode. For a fast intra coding unit (CU) size decision, we predict the RD cost of four split sub-CUs from the RD of the current CU using the relationship between the RDs of the current CU and its split sub-CUs. In the HEVC Test Model software, the proposed algorithm saves 48.56% of the encoding time, on average, while the RD loss is only 0.59%. In x265, the proposed algorithm improves the encoding speed to 71.95 fps, on average, from 49.46 fps. It achieves 1.8 times faster speed with a bit increase of only 0.44%, compared to x265.

The High Efficiency Video Coding standard and its screen content coding extension provide superior coding efficiency compared to predecessor standards. However, this coding efficiency is achieved at the expense of very complex encoders. One major complexity driver is the comprehensive rate distortion (RD) optimization. In this paper, we present a deep learning-based encoder control which replaces the conventional RD optimization for the intra prediction mode with deep convolutional neural network (CNN) classifiers. Thereby, we save the RD optimization complexity. Our classifiers operate independently of any encoder decisions and reconstructed sample values. Thus, no additional systematic latency is introduced. Furthermore, the loss in coding efficiency is negligible with an average value of 0.52% over HM-16.6+SCM-5.2.

This study proposes a LCU (Large Coding Unit) level fast algorithm of video coding based on AVS3. Nowadays, AVS3 (Audio Video coding Standard 3) has implemented traversal algorithm on CU (Coding Unit) prediction modes. It first computes the current blocks RD cost of inter-frames mode and then compute the RD (Rate Distortion) cost of intra-frames mode. By comparing RD cost, it chooses the best prediction mode. Inter-frames modes include three modes: skip, direct and inter. While intra-frames modes include two: intra and IBC (Intra Block Copy). This method will run different prediction modes algorithm when it encodes one CU and can choose the optimal prediction mode, though with large computation complexity. Based on previous researches, we do more surveys and expand the LCU of AVS3 from 64*64 to 128*128 and add more prediction tools. Therefore, the prediction mode in space express stronger correlation and we present a fast prediction mode algorithm based on apposition LCU. According to apposition LCU image content of adjacent frames in the same video and correlation with optimal prediction mode, we can provide prediction mode reference for the LCU not encoded and reduce the number of traversals. The time complexity is decreased significantly. In our experiment, it shows that under the LDP (Low delay P-frame) model, twelve general test cases reduce 16.67% in time complexity and the performance loss is only 0.61%. With fewer performance loss, we dramatically reduce time complexity and the experiments results is overall satisfying.

In this paper, we study the rate-distortion (RD) optimization of the H.264 main profile encoding. Specifically, a soft decision quantization (SDQ) algorithm is developed based on the context adaptive binary arithmetic coding (CABAC) method in the H.264 main profile. Given motion prediction and quantization step sizes, the proposed SDQ algorithm is proved to achieve near-optimal SDQ for residual coding in the sense of minimizing the true RD cost when the weak adjacent block dependency utilized in CABAC is ignored for optimization. The SDQ algorithm is then used in conjunction with a general RD optimization framework to jointly design motion prediction and residual coding for H.264 main profile coding given previously coded reference frames. Experiments have been conducted based on the reference encoder JM82 of H.264 main profile. Comparative studies show that the joint design method achieves on average 10% rate reduction at the same PSNR when compared with the RD method in the H.264 main-profile reference software, with half of the reduction coming from the proposed SDQ algorithm, and 20% rate reduction at the same PSNR when compared with the RD method in the H.264 baseline-profile reference software.

Hybrid multiple description video coding using optimal DCT coefficient splitting and SD/MD switching