This paper presents an object-based layered video coding scheme which achieves very high compression efficiency along with the provision for advanced content-based functionalities, e.g., content-based scalability or content-based access and manipulation of video data. In a first step, a video sequence is segmented into several arbitrarily shaped "object layers." To achieve the desired content-based functionalities, a baseline shape-adaptive discrete cosine transform (DCT) coding algorithm is introduced which can be seen as an extension of conventional block-based DCT coding schemes (e.g., H.261, H.263, MPEG-1, or MPEG-2) toward coding of arbitrarily shaped image content. In order to increase compression efficiency, the baseline object-based layered approach can be extended with an object-based motion prediction toolbox. Using this toolbox, the coding scheme can potentially select specific prediction techniques for every object layer to be coded. To illustrate the concept, an extension of the baseline shape-adaptive DCT algorithm with a technique for global background motion estimation and compensation is described which significantly improves the compression efficiency of suitable video sequences compared to standard MPEG coding schemes.

This contribution represents the results of subjective tests organized in November '94 by the HAMLET subgroup WP2. These tests, carried out in accordance with Rec. ITU-R 500-5, compare the performances of different types of MPEG-2 encoders: (1) the software encoder of the MPEG-2 technical documentation (reference); (2) an exact simulation of the hardware encoder that is being built in the same RACE project, (3) an optimized software encoder and (4) an SNR scalable encoder. The results reveal that the MPEG-2 standard leaves much room for optimization in the encoder. They also illustrate that optimizations carried out for the HAMLET hardware encoder lead to important improvements. Finally, they show the almost comparable quality of the MPEG-2 reference software encoder at 7 Mbit/s and the SNR scalable encoder at 3+4 Mbit/s for the base + enhancement layer.

For a feasible calculation of filters for the subband coding of picture sequences with linequincunx subsampling, a coordinate transformation of the sampling raster is proposed. With the new coordinates, the linequincunx subsampling becomes a simple quincunx subsampling without interlace. This facilitates the mathematical derivation of three-dimensional perfect reconstruction filters.>

In the above named work by B. Girod et al. (see ibid., vol.36, no.3, p.332-46, 1988), a stability condition for a DPCM (differential pulse code modulation) system with additional quantization error noise shaping has been proposed. In the present work, a correction of this stability condition is derived.>

In the above-named work (see ibid., vol.IT-33, p.438-43, May 1987), a method was presented for the construction of a variable length code with maximum word-length constraint. The algorithm is partially based on the algorithm of A.H. Murakami et al. (1984), and is extended with heuristic choices for the bifurcations. C.H. Lu (1988) explains why step (1) taken by Murakami, which is used in De Lameillieure's contribution as condition (6), does not guarantee an optimum code when a word-length restriction is imposed. This step (1), or condition (6), does not take the bounded maximum word length into account. Therefore, in the present contribution the author rectifies this difficulty by proposing an adaptation of condition (6) into a composite condition for the case when there is a word-length constraint. Proofs of the components of this composite condition are presented together with numerical results.>

A depth-first algorithm is presented for the construction of a binary minimum-redundancy variable length code with limited word length. In this algorithm, heuristic information on the mean word length is used for efficient searching. The extension toQ-ary codes is also discussed.

In the above paper, conditions are derived for determining whether or not a scheme with transmission of predictor information for selected pels is better than a scheme with adaptive predictor switching only. These comments contain experimental evidence which proves that those conditions are not valid. Moreover, it is pointed out which part of the derivation of the above conditions is erroneous.

In the above correspondence, two algorithms for the construction of a minimum redundancy code have been described. In this correspondence, it is pointed out that the strategy in the correspondence does not always lead to an optimal code. A correction of both algorithms is presented.