This paper is devoted to the design and the steady-state performance analysis of a data-aided or decision-directed timing recovery scheme for filter bank-based multicarrier transmission with equalization. The symbol synchronizer is obtained from a Mueller and Muller approach extended to multiband transmission. A simple synchronizer is first proposed which can be derived from an approximate likelihood function. Then, possible improvements are discussed. Timing jitter variances are computed and discussed for high bit-rate transmission over copper wires.

The performance degradation of filter-bank-based multicarrier transmission due to timing errors is investigated. The receiver is made of a fractionally spaced linear or decision-feedback equalizer designed for some sampling phase. The actual sampling phase is different and the impact of the difference on the performance is investigated. Sampling phase offset and jitter are considered. Besides, assuming the sampling phase error can be estimated the efficiency of various types of interpolation is investigated.

This paper examines a decision-directed timing recovery scheme for filter-bank-based multicarrier transmission. The symbol synchronizer is obtained from a maximum-likelihood approach and uses the digital equalization filters. The algorithm is first derived, the performances of the scheme (timing offset and timing jitter) are then computed, and computational results are provided for high bit-rate transmission over copper wires. Finally, different improvements are discussed.

This paper is devoted to a decision-directed timing recovery scheme for filter bank based multicarrier transmission. The symbol synchronizer is obtained from a maximum-likelihood approach and uses the equalizer outputs. The algorithm is first derived and computational results for the resulting timing error are provided for high bit rate transmission over copper wires. Finally, different improvements are discussed.

In this paper, we investigate the statistical properties of interlaced and progressive image sequences in order to determine what is the most efficient format from a coding point of view. Intra- and inter-coding modes are both considered. The main result is that, for both intra- and inter-images, the coding of progressive images outperforms the coding of interlaced ones for low bit rates and large motion vectors. Particular attention is paid to inter-coding with subpel accuracy, where the effects of nonideal interpolation filters and finite precision of the motion vector are taken into account.

We investigate detectors for transmultiplexer-based high bit rate transmission over dispersive channels like telephone wires. The transmitter is based on a synthesis stage. At the receiver side, we design, for an MMSE criterion, multiple-input/multiple-output (MIMO) linear detectors applied after an analysis stage and upgraded with a MIMO decision-feedback (DF) section. The performance of the system is investigated for filter banks over copper wires with different noises (AWGN, FEXT and RFI). The influence of the forward section rate is investigated. The potential of these detectors is clearly demonstrated. Besides, reduced complexity solutions are also put forward.

We investigate the asymptotic performance of a multiple input/multiple output (MIMO) decision-feedback equalizer (DFE) used to detect a multicarrier signal transmitted without guard time over a linear dispersive channel. We design the MIMO DFE for a minimum mean square error criterion (MMSE) and compare its performance to that of a single carrier system also using an infinite length MMSE DFE equalizer. We demonstrate that both uncoded systems achieve the same capacity.

This correspondence addresses the problem of human visual weighted quantization for transform/subband coding of interlaced pictures. Merged field coding is assumed. The criterion proposed for the optimization of the quantizer is the distortion measured on the deinterlaced and decoded image. This criterion and the associated weighting factors are shown to be dependent on the motion between two successive fields.

This paper presents a theoretical analysis of the coding efficiency of interlaced and progressive formats for intra and prediction error images. It shows that progressive coding is more effective for lower bitrates and larger motion vectors than interlaced coding.

The generalized sampling theorem states that any analogue signal whose spectrum is limited to 1/T can be exactly recovered from N sequences of samples taken at a rate 2/NT and all having a different sampling phase. When N=2, the exact interpolation formula can be derived quite easily. The ideal interpolation filters have infinite impulse responses. This paper addresses the theoretical question of recovering from the 2 initial sequences, any other sequence taken at the some rate 1/T and with a different sampling phase. FIR filters are derived for a least square interpolation error. Moreover, an adaptive implementation is proposed and formulated as a Kalman algorithm. Simulation results obtained for AR processes show the effectiveness of the solution compared to static solutions.

Delogne et al. (see IEEE Transactions on Image Processing, vol.3, no.5, p.482-91, 1994) have shown how the motion estimation and compensation have to be carried out when interlaced images are processed under the assumption of a translational motion. Due to the finite accuracy of the motion vectors, the prediction error is nonzero after motion compensation. We therefore investigate the influence of quantizing the motion vectors on the statistical properties of the prediction error signal. The power spectral density and the variance of the prediction error are also computed.

The generalized sampling theorem states that any analogue signal whose spectrum is limited to 1/T can be exactly recovered from N sequences of samples taken at a rate 2/NT and all having a different sampling phase. When N=2, the exact interpolation formula can be derived quite easily. The ideal interpolation filters have infinite impulse responses. This paper addresses the question of recovering from the 2 initial sequences, any other sequence taken at the same rate 1/T and with a different sampling phase. The design problem is dealt with for finite length filters and the criterion is the minimization of the mean squared interpolation error.>

The paper addresses the problem of human visual weighted quantization for transform/subband coding of interlaced pictures. This problem has been correctly solved for progressively scanned pictures [Vandendorpe 1991]. When interlaced pictures are coded, the problem can be solved easily for field coding. However, if the coding of merged fields (known as frame coding) is preferred, the quantizer optimization becomes more difficult. In the paper, the authors show how to solve the problem for merged field coding by using the motion information which exists between 2 successive fields. The weighting factors required to determine the quantization step sizes of the subbands are shown to be dependent on this motion information. In addition, the variances of the subbands produced after held merging are also dependent on the motion. Their exact expression is derived.>

This paper investigates a coding method for interlaced image sequences which uses deinterlaced images as an intermediate format. The purpose of this method is threefold: firstly, to achieve a higher coding gain, as the processing of progressive pictures raises much less difficulty than the processing of interlaced ones; secondly, to provide an intermediate step toward the development of a fully progressive production and transmission chain; and finally to open the way to efficient solutions for important image processing issues such as frame rate conversion and compatibility. In previous work by the authors (1994), an interlace-to-progressive converter was proposed. It was based on theoretically correct generalized interpolation formulas, working under the assumption of a uniform translational motion. The scheme proposed here makes use of this converter to produce the progressive images from the interlaced source, which are coded and then re-interlaced. Another solution is considered, with the purpose of improving the coding efficiency alone. It is based on an intermediate progressive format at half the original temporal sampling rate. However, in that case, some additional information needs to be transmitted in order to accurately recover the full rate interlaced signal. The two methods mentioned above are described in some detail and tested on actual sequences. The picture quality is assessed, both objectively and subjectively, and conclusions are drawn.>

This paper introduces a new method for the interpolation that has to be performed when motion estimation and compensation are applied to interlaced sequences with subpel accuracy. It is based on the assumption that a uniform motion exists between two successive frames. The exact formulas for the estimation are derived. They show that in order to obtain a correct interpolation of each field of one frame, it is necessary to use the information of both fields of this frame. Because the ideal filters have infinite impulse responses, the filter design is discussed, and the efficiency is measured for typical sequences.

This paper deals with the question of optimizing the filters in the upsampling stage of a TV/HDTV compatible pyramidal coder. From a coding gain point of view, both the decimation and upsampling filters should be optimized. In the frame of compatible coding, not only the coding efficiency influences the choice of the decimation filter but also the compatible image quality. Therefore, assuming this filter has been fixed, we analyze the question of optimizing the upsampling filter in order to obtain the highest coding gain. This question is addressed for a mean squared error (MSE) criterion. In addition, assuming the base layer (TV) signal can be quantized, the influence on the quantization noise on the optimal interpolation filter is investigated and the problem is handled for the MSE criterion. As the statistical properties of pictures are required in the optimization, a model is then developed to compute these properties when there is motion. The model takes into account the processing of progressive sources and, concerning interlaced sequences, the independent processing of fields or the processing of merged fields. Results are then derived for the three types of processing.

This paper deals with the problem of motion estimation and compensation with sub-pixel accuracy when interlaced moving sequences are processed. Provided a translational motion exists between consecutive frames (or that the image can be partitioned into sets of points with such a motion) the exact formulas for the prediction of pixels are derived. It is shown that in order to obtain a correct prediction of any pixel in one frame, it is required to use the information of both fields of the previous frame. The ideal interpolation filters are of infinite length. Therefore, the question of designing finite length filters is addressed and alternative design methods are proposed. The efficiency of the method is then measured on typical sequences.