Search

We describe a multiple description (MD) video coding technique using 3D-Set Partitioning in Hierarchical Tree algorithm (3D-SPIHT). Multiple description coding (MDC) is a way of coding video information into multiple bitstreams and is useful in a number of scenarios including peer-to-peer video streaming and video transmission over error-prone networks. We propose a modified tree structure for 3D-SPIHT that is more efficient for MD coding. We also propose a branch-pruning technique to generate multiple descriptions. We then show how rate-distortion optimization can be incorporated into the MDC framework to develop an efficient MD video coding system. Experimental results show that the proposed MD system achieves significant improvement in performance compared to existing methods. Index Terms--Branch pruning, multiple description coding (MDC), redundancy, set partitioning, tree structure.

Shape masks are used in object-based video coding, such as MPEG-4 part 2, to specify the boundaries of a video object. The term 'binary shape' is used to describe shape masks where each pixel is either completely inside the object or completely outside it, i.e., there is no blending of pixels at object boundaries. In this paper, we present a new lossless block-based coding algorithm for binary shapes that combines a quad-tree structure with context-based arithmetic coding. Experimental results demonstrate that this new approach provides a saving in bits generated up to approximately 66% compared to the MPEG-4 binary shape coding algorithm. Index Terms--Context-based arithmetic encoding (CAE), quad-tree, shape coding.

An analysis on the reduction of the so-called boundary artifacts that occur at the edges of reconstructed images when wavelet transforms are used to perform lossy image compression is presented. The odd tile length low-pass first (OTLPF) convention provides a simple method to significantly reduce coding artifacts at tile boundaries in wavelet-based image compression.

In many scenarios where it is desirable to provide a number of different descriptions of a video sequence, the simulcasting of single-layer bitstreams is impractical due to the amount of data involved. The new technique 'stream morphing' introduced in this paper exploits the redundancy that exists between different compressed descriptions of the same video sequence at different quality levels. This works to dramatically reduce the data rates required to achieve the same effect as simulcast. The results shown here indicate that an SNR scalable video system constructed using this principle outperforms the structures defined in MPEG-2 and the MPEG-4 Streaming Video Profile. Stream morphing has a number of additional advantages: no loss of subjective quality compared to single-layer bitstreams at the same peak SNR regardless of the number of layers involved (unlike the MPEG-2 approach), the ability to generate a scalable representation of a video sequence from a set of single-layer descriptions in a computationally inexpensive way, and compatibility with existing single-layer decoders via a simple preprocessor without the need for transcoding. Stream morphing is demonstrated here in the context of MPEG-4: however, it is compatible with other similar video coding standards. Index Terms--Fine granularity scalability (FGS), layered video coding, SNR scalability, stream morphing.

This project was designed to examine university graduates' expectations and experiences of employment in small organizations in the UK and the Netherlands. Specifically, three predictions were made on the basis [...]

It is well known that layered video coding is useful both for service interworking and as an aid to error resilience. A major drawback of layered coding is that it invariably results in a reduction in overall coding efficiency for the high quality service. An attractive approach would be one in which the high quality service is coded, and then the resulting bitstream is decomposed in such a way that the lower resolution services can be reconstructed using only a subset of the total generated data. This would mean that there would be no impact on the coding efficiency of the highest quality service. In this paper we demonstrate that the quality of the lower layer when this approach is used is fundamentally limited by drift. It is shown that even a coarsely quantized, low rate correction signal provides a major benefit to the quality of this lower layer service while still not impacting significantly on the coding efficiency of the high quality service. Of course, some overhead is introduced in the coding of the lower quality service but the total overhead is still significantly better than simulcast. Index Terms - Data partitioning, frequency scalability, layered coding, MPEG, video coding.

This paper presents a new fast search algorithm for block matching motion estimation called the adaptive search length (ASL) algorithm. The ASL algorithm adaptively varies the number of positions searched for each block while still maintaining control of the average number of searches per block for each frame. Experimental results show that the peak signal-to-noise ratio (PSNR) of decoded sequences which were coded using the ASL algorithm is within 0.25 dB of the PSNR of decoded sequences which were coded using the full search block matching algorithm. It is also shown that the ASL algorithm requires only 10% of the computations required by the full search algorithm to achieve this level of decoded image quality. Index Terms - Image coding, motion compensation.

The structure of hyperspectral images reveals spectral responses that would seem ideal candidates for compression by vector quantization. This paper outlines the results of an investigation of lossless vector quantization of 224-band Airborne/Visible Infrared Imaging Spectrometer (AVIRIS) images. Various vector formation techniques are identified and suitable quantization parameters are investigated. A new technique, mean-normalized vector quantization (M-NVQ), is proposed which produces compression performances approaching the theoretical minimum compressed image entropy of 5 bits/pixel. Images are compressed from original image entropies of between 8.28 and 10.89 bits/pixel to between 4.83 and 5.90 bits/pixel.

The outline of new course in video coding is presented in which the basic concepts are reinforced by way of a series of graded MATLAB exercises. Following a brief overview of the motion-compensated discrete cosine transform (DCT) encoder, a number of exercises are described in which the student obtains an understanding of image correlations and entropy and the concepts involved in predictive encoding. The data compaction properties of the discrete cosine transform are then illustrated together with interframe motion-estimation and motion-compensated prediction. All aspects of the evolving video encoder are then integrated by way of a simplified encoder syntax that is based upon the standard Moving Picture Experts Group (MPEG) syntax. This allows students to exchange digital bit-streams and decode image sequences that have been encoded by fellow students. Advanced topics such as fast motion estimation techniques, error resilience, and other coding approaches are briefly discussed.

The explicit finite difference beam propagation method is employed to solve the three-dimensional coupled wave equations used in modeling the nonlinear behavior of optical waveguides. The method is implemented on a transputer array and a connection machine for parallel processing. Results indicate that the new parallel algorithm is very useful for modeling complicated nonlinear optical devices.

A new rate control algorithm is developed for a variable bit-rate (VBR) video coder. The quantizer step size of the coder is changed by the algorithm, the changes being influenced by the image sequence that affects the monitoring of errors. The output bit-rate of the coder is limited by the algorithm without the requirement of a buffer. Decoded image sequences exhibit a high quality, as the VBR encoder uses the new algorithm which does not require a buffer for the coder limitation.

An examination of the distribution of signals within the prediction loop of an interframe video codec reveals that small random fluctuations caused by the quantizer build up over a period of time. This quantization noise circulates within the prediction loop causing a gradual decrease in the signal to noise ratio of the reconstructed sequence and an increase in the required data rate. In general, both of these effects can be nullified by the inclusion of a low pass loop filter.

Reversible image compression rarely achieves compression ratios larger than about 3 : 1. An explanation of this limit is offered, which hinges upon the additive noise the sensor introduces into the image. Simple models of this noise allow lower bounds on the bit rate to be estimated from sensor noise parameters rather than from ensembles of typical images. The model predicts that an 8-b single-band image subject to noise with unit standard deviation can be compressed reversibly to no less than 2.0 b/pixel, equivalent to a maximum compression ratio of about 4 : 1. The model has been extended to multispectral imagery. The Airborne Visible and Infra Red Imaging Spectrometer (AVIRIS) is used as an example, as the noise in its 224 bands is well characterized. The model predicts a lower bound on the bit rate for the compressed data of about 5.5 b/pixel when a single codebook is used to encode all the bands. A separate codebook for each band (i.e., 224 codebooks) reduces this bound by 0.5 b/pixel to about 5.0 b/pixel, but 90% of this reduction is provided by only four codebooks. Empirical results corroborate these theoretical predictions.

An extensive three-year project was conducted to answer questions on the career development of college graduates in the UK. The subjects worked in financial services, food, engineering, chemicals, distribution and pharmaceuticals industries. A majority of the subjects opted to work with their employers because of long-term career and training opportunities. The graduates were not trying to get other jobs, but were neither entirely committed to working with their current employers.

The negative portrayal of blacks as looters for doing the exact same things as whites is elaborated. Media portrayal of the behaviors of African Americans during the devasting hurricanes, which presents disjointed patterns of people struggling to strive against nature, is described.

In the real world, we commonly receive information simultaneously through two or more senses, with the brain fusing this data to produce a single coherent message. Lip-reading is one example of this phenomenon. Laboratory studies, on the other hand, often measure the response to a stimulus by a single sense and extrapolate these results to predict real-world behavior. In this paper, we show that semantics have a significant impact on viewers' sensitivity to the quality of a video sequence for spatially separated parts of the sequence and, more importantly, that this difference in sensitivity can be changed by the presence of an audio signal. This result is important for any testing of subjects' responses to visual material. One example is the subjective assessment of the quality of video in an audio--visual communications system (such as television or videoconferencing). Index Terms--Joint audio--visual coding, joint audio--visual subjective testing, video coding, video subjective quality, video subjective testing, videoconferencing.

The study of the transmission of electromagnetic waves through a photonic crystal with various membranes placed over the surface is presented in this paper. A difference in performance is observed even for a membrane thickness that is unable to support guided substrate modes through total internal reflection. The transmittance has been investigated for two crystal orientations, assuming normally incident external plane waves on a finite thickness two-dimensional (2-D) photonic crystals both with and without a membrane. The angular transmission response is characterized by scanning the incidence angle of the impinging plane wave to cover all available angles within the 2-D periodic plane of the structure. Index Terms--Antennas, electromagnetic crystals, periodic structures, photonic bandgap crystals.

We describe a robust codec for the transmission of very low bit-rate video over channels with a variety of errors, including random and bursty bit errors and packet loss. The codec exploits adaptivity to give good performance with a low overhead. By only protecting macroblocks which would otherwise be poorly concealed by the decoder, the codec allows adaptive selection of the parts of video to protect. For protection, it uses multiple description codes which indirectly provide frequency-based adaptivity by protecting the more significant DCT coefficients. Simulations show significant improvements in the performance of the codec when compared to codecs which use intra macroblock updating (raster scan and random) at the same overhead. The codec is efficient in its use of bits and has good error resilience properties both objectively and subjectively over a wide range of conditions. Further, transcoding of the received bit stream to the standard H.263 syntax is relatively easy. Index Terms--Error resilience, multiple description codes, robust video transmission, unequal error protection.

Audio-visual and other multimedia services are seen as important sources of traffic for future telecommunication networks, including wireless networks. A major drawback with some wireless networks is that they introduce a significant number of transmission errors into the digital bitstream. For video, such errors can have the effect of degrading the quality of service to the point where it is unusable. In this paper, we introduce a technique that allows for the concealment of the impact of these errors. Our work is based on MPEG-2 encoded video transmitted over a wireless network whose data structures are similar to those of asynchronous transfer mode (ATM) networks. Our simulations include the impact of the MPEG-2 systems layer and cover cell-loss rates up to 5%. This is substantially higher than those that have been discussed in the literature up to this time. We demonstrate that our new approach can significantly increase received video quality, but at the cost of a considerable computational overhead. We then extend our technique to allow for higher computational efficiency and demonstrate that a significant quality improvement is still possible. Index Terms--Author: please supply index terms. E-mail: keywords@ieee.org for information.

Signal-to-noise ratio (SNR) scalability has been incorporated into the MPEG-2 video-coding standard to allow for the delivery of two services with the same spatial and temporal resolution but different levels of quality. In this paper, we begin by reviewing the performance of a single-loop SNR scalable encoder that is compliant with the MPEG-2 standard and demonstrate that its performance is limited by drift in the base layer. We also look at an alternative single-loop drift-free noncompliant SNR scalable encoder, but discover that its coding efficiency is poor. We then review the performance of an MPEG-compliant two-loop SNR scalable encoder. Finally, we propose a new two-loop noncompliant encoder which achieves improved coding performance at the expense of some increase in encoder and decoder complexity. Index Terms--Layered coding, MPEG-2, SNR scalability, video coding.