Yafan Zhao

Low-complexity skip prediction for H.264 through Lagrangian cost estimation

A complexity reduction algorithm for an H.264 encoder is proposed. Computational savings are achieved by identifying, prior to motion estimation, macroblocks (MBs) that are likely to be skipped and hence saving further computational processing of these MBs. This early prediction is made by estimating a Lagrangian rate-distortion cost function which incorporates an adaptive model for the Lagrange multiplier parameter based on local sequence statistics. Simulation results demonstrate that the algorithm can achieve computational savings of 19%-67% (depending on the source sequence) with no significant loss of rate-distortion performance.

Complexity Control of H.264/AVC Based on Mode-Conditional Cost Probability Distributions

A computational complexity control algorithm is proposed for an H.264 encoder running on a processor/power constrained platform. This new computational complexity control algorithm is based on a macroblock mode prediction algorithm that employs a Bayesian framework for accurate early skip decision. Complexity control is achieved by relaxing the Bayesian maximum-likelihood (ML) criterion in order to match the mode decision threshold to a target complexity level. A feedback algorithm is used to maintain the performance of the algorithm with respect to achieving an average target complexity level, reducing frame by frame complexity variance and optimizing rate-distortion performance. Experimental results show that this algorithm can effectively control the encoding computational complexity while maintaining a good rate-distortion performance at a range of target complexity levels.

A Full Reference Quality Metric for Compressed Video Based on Mean Squared Error and Video Content

Visual quality of compressed video sequences depends on factors including spatial texture content and cognition-based factors such as prior knowledge and task in hand. The MOSp metric is a full reference objective quality metric which predicts perceived quality of sequences with video compression-induced impairments based on the spatial texture content and the mean squared error between original and compressed video sequences. In this paper, we extend the MOSp metric to incorporate cognition-based factors to identify regions in a video scene that attract human attention. The proposed metric has been tested on a variety of multimedia sequences of common intermediate format resolution compressed at a wide range of bitrates using the H.264/AVC coding standard. This metric shows a higher correlation with mean opinion score (MOS) than popular metrics, such as peak signal-to noise ratio, National Telecommunications and Information Administration/Institute for Telecommunication Sciences video quality metric, PSNRplus, and the Yonsei University metric. Results also show that by extending the MOSp metric to incorporate cognition-based factors such as skin information, its correlation with subjective scores (MOS) can be significantly improved in video content where humans are present. This algorithm is particularly useful for real-time quality estimation of multimedia sequences with block-based video compression-induced impairments because all the parameters of the metric can be calculated automatically with a modest amount of processing overhead.

Computational complexity management of motion estimation in video encoders

Summary form only given. The performance of software-only video codecs is often constrained by available processing power. Existing fast motion estimation algorithms are not designed to provide flexible, predictable control of computational complexity. We propose an adaptive algorithm, which maintains the computational complexity of the motion estimation function at various target levels by controlling the motion estimation search pattern.

Adaptive algorithms for variable-complexity video coding

Variable-complexity algorithms provide a means of managing the computational complexity of a software video CODEC. The reduction in computational complexity provided by existing variable-complexity algorithms depends on the video scene characteristics and is difficult to predict. A new approach to variable-complexity encoding is proposed. A variable-complexity DCT algorithm is adaptively updated in order to maintain a near-constant computational complexity. The adaptive update algorithm is shown to be capable of providing a significant, predictable, reduction in computational complexity with only a small loss of video quality. The proposed approach may be particularly useful for software-only video encoding, in applications where processing resources are limited.

Complexity management for video encoders

Computational complexity is an important performance constraint for software-only video CODECs. The aim of this research is to develop a video coding system with variable, controllable computational complexity. Adaptive algorithms for DCT and motion estimation are proposed separately to reduce complexity of each function and maintain it at target level. An integrated approach to video CODEC complexity management is also addressed. This work will have potential benefit for a wide range of computation-constrained or power-constrained multimedia applications.

Adaptive Management of Video Encoder Complexity

Abstract In this paper, we investigate methods of reducing the computational complexity of the discrete cosine transform (DCT) in a software video encoder. The number of DCT calculations may be reduced by modeling the distribution of zero blocks. We demonstrate that the reduction in computational complexity is variable and depends on the statistics of the video sequence. We propose a new adaptive algorithm that can maintain a near-constant reduction in complexity. The proposed algorithm performs well at converging to a “target” computational complexity, at the expense of a small reduction in image quality. This algorithm provides a flexible mechanism for managing computational complexity in a video encoder.

Macroblock classification for complexity management of video encoders

Typically, many macroblocks (MBs) are skipped during encoding of H.263 or MPEG-4 SP video data, particularly at low bit-rates. In this paper, we describe an algorithm that predicts the occurrence of skipped MBs prior to encoding, making it possible to save significant computational effort by not coding these MBs. The algorithm estimates the energy of low-frequency quantized coefficients in order to classify each MB as ‘skipped’ or ‘not skipped’. Results show that the algorithm can deliver substantial computational savings at the expense of a small reduction in rate-distortion performance.

A MAP Framework for Efficient Skip/Code Mode Decision in H.264

While the H.264 standard offers improved compression efficiency compared with prior video coding standards, this efficiency arises at the cost of significant complexity. We present a reduced complexity coding algorithm which estimates, prior to coding each macroblock, whether the coder would choose the skip or code mode for the macroblock. Computation savings are achieved, since pre-coding of skipped macroblocks is avoided. The decision to skip a macroblock is based on estimating and modelling mode cost differences, and employing these models in a MAP framework. Results are shown which indicate that for low-activity sequences savings of over 70% in computation time can be achieved with little or no decrease in video quality.