Guido Heising

Video compression using context-based adaptive arithmetic coding

A new entropy coding scheme for video compression is presented. Context models are utilized for efficient prediction of the coding symbols. A novel binary adaptive arithmetic coding technique is employed to match the conditional entropy of the coding symbols given the context model estimates. The adaptation is also employed to keep track of non-stationary symbol statistics. Our new approach has been integrated into the current ITU-T H.26L test model (TML) to demonstrate the performance gain. By using our new entropy coding scheme instead of the variable length code approach of the current TML, large bit-rate savings up to 32% can be achieved. As a remarkable outcome of our experiments, we observed that high gains are reached not only at high bit-rates, but also at very low rates.

Context-based adaptive binary arithmetic coding in JVT/H.26L

In this paper a new adaptive entropy coding scheme for video compression is presented. It utilizes an adaptive arithmetic coding technique to better match the first order entropy of the coded symbols and to keep track of nonstationary symbol statistics. In addition, remaining symbol redundancies are exploited by context modeling to further reduce the bit-rate. A novel approach for coding of transform coefficients and a table look-up method for probability estimation and arithmetic coding is presented. Our new approach has been integrated in the current JVT test model (JM) to demonstrate the performance gain, and it was adopted as a part of the current JVT/H.26L draft.

Video coding using spatial extrapolation based motion field segmentation

Temporal image prediction using motion compensation techniques has been used successfully to encode digital image sequences. In current standards (MPEG and H.263) block matching is used for motion estimation and only translational motion and motion discontinuities (MD) at block edges can be described. To overcome this problem the spatial-transform was introduced. Using affine, geometric transforms these schemes are capable of describing rotation, shear, and change in scale. Due to the underlying continuous motion model problems arise at the MDs. We propose a two step local MD estimation scheme. In the first step motion estimation is performed, in the second step the MDs are computed for every block. The dense motion fields on both sides of a discontinuity line are extrapolated from their respective neighbours. The selection of the optimal position of the MD yields a higher coding efficiency and an improved visual quality. Although the MDs have to be transmitted as additional side information a coding gain is achieved compared to conventional coders.

Progressive texture video coding

We present a novel video coding approach which provides the functionality of fine granular bitstream scalability. The proposed progressive texture video coding (PTVC) scheme consists of two parts: the current ITU-T H.26L test model for coding macroblock mode and motion information only; a new technique for progressively coding all intra and inter texture information thus providing an embedded bitstream. It can still be decoded with moderate quality degradation in case of partial loss of the bitstream. It is therefore well suited for multicast or broadcast scenarios where time varying or different fixed bitrates are provided to the clients, as no transcoding or multiple coding process is required.

Efficient fine granular scalable video coding

We present an efficient fine granular scalable video compression scheme which supports a fast bit rate adaptation independent of the encoder. The proposed scheme generates an embedded bitstream for each frame or, by appropriate multiplexing, for each group of picture. This rate-scalability is supported by an embedded bitstream which allows decoding at multiple rates, or to be more specific at virtually any rate. Drift removing techniques based on intra refresh or feedback mechanisms are presented. We show the potential of this video codec for variable bit rate channels as well as in combination with an unequal erasure protection scheme for error robust and efficient transmission over packet erasure channels. Possible enhancements of the presented coding schemes are discussed.

Video coding using a bilinear image warping motion model and wavelet-based residual coding

This paper describes a video coding algorithm that combines new ideas in motion estimation, wavelet filter design, and wavelet-based coding techniques. A motion compensation technique using image warping and overlapped block motion compensation is employed to reduce temporal redundancies in a given image sequence. This combined motion model has the advantage of representing more complex motion than simple block matching schemes. Spatial decorrelation of the motion compensated residual images is performed using an one- parametric family of biorthogonal IIR wavelet filters coupled with a highly efficient pre-coding scheme. Experimental results demonstrate substantial improvements in objective quality of 1.0 - 2.2 dB PSNR compared to the H.263+ standard. Especially at very low bit-rates where the reconstruction quality of block-based coders suffers from visually annoying blocking artifacts the proposed coding scheme produces a superior subjective quality.

Efficient and robust motion estimation in grid-based hybrid video coding schemes

In the past advanced motion compensation techniques have been introduced leading to improved temporal predictions in hybrid video coding schemes. The focus of this paper is on grid-based warping prediction. The advantage of this technique compared to the commonly used block translational motion compensation is its improved capability to handle non-translational motion. However, a disadvantage is its 2-dimensional interdependency of motion vectors. This can lead to poor predictions, if the dependencies are not considered during the motion estimation process. Therefore in this paper a new highly efficient motion estimation method for grid-based prediction is presented. Different means are developed to improve the prediction quality, to decrease the computational complexity and to enhance the robustness of the algorithm. By introducing a hierarchical iterative octagonal matching with spatial and temporal prediction of candidate vectors for the motion search into a wavelet-based hybrid video coder a coding gain of up to 1.5 dB compared to MPEG-4 can be achieved.

Improving coding efficiency in grid-based hybrid video coding schemes

In the past many grid-based video coding schemes using image warping p17ediction have been proposed. The advantage of grid-based motion compensation is its ability to model non-translational motion. Disadvantages are the interdependency of motion vectors which require more complex motion estimation strategies and its restricted ability to model motion boundaries. In this paper different solutions, to cope with the latter deficiency are presented and compared with respect to coding efficiency. In particular, three adaptive modes for handling motion discontinuities and a block-adaptive loop filtering are presented. It is shown that these are key components for achieving a high coding gain in grid-based hybrid coding. Introducing these features into a wavelet-based video coder leads to a coding gain of up to 2,5 dB compared to ITU H263 + and MPEG-4.

Blocking artefact free video coding based on a bilinear forward image warping model

In this paper a new very low bit rate video coding scheme is presented. It is based on a forward motion estimation technique using an image warping model, thus delivering the basis for temporal tracking of regions or objects. In order to deal with new or disappearing image contents a new algorithm for the adaptation of a tracked control grid is proposed. To improve the coding efficiency a high quality pixel interpolation in the warping based predictor has to be used, which unfortunately also emphasizes the high frequency coding artefacts. To reduce these artefacts the prediction error is coded by a biorthogonal wavelet transform. The proposed coder yields an improvement of the PSNR of ca. 0.7 dB compared to the advanced H.263 coder. In addition the visual quality is better as any blocking artefacts in the coded images are avoided.

A wavelet-based video coding scheme using image warping prediction

This paper describes a video coding algorithm that combines new ideas in motion estimation and wavelet-based coding techniques. A motion estimation technique using an image warping model is employed to reduce temporal redundancies in a given image sequence. The underlying continuous motion model is based on a bilinear warp of a grid of quadrangles. It has the advantage of dealing with more complex motion than simple block matching schemes. In addition, a block adaptive high-quality texture interpolation is used to further improve the quality of the prediction. The prediction error image is decorrelated using a wavelet transform coupled with a framework of highly efficient pre-coding techniques utilizing the concepts of partitioning, aggregation and conditional coding (PACC). This combined coding strategy has proved to be very efficient both with respect to subjective and objective measures. The improvements achieved with our proposed coder compared to the H.263 coder (with all options applied) range from 0.5 to 1.2 dB PSNR. Especially at very low bit-rates where reconstructed videos of block-based coders suffer from visually annoying blocking artifacts the proposed coding scheme produces a superior subjective quality.