This paper reviews some of the latest developments in image compression for digital transmission of HDTV. Besides a description of the coding schemes of already implemented systems or systems under development, the paper focuses on contribution and distribution aspects of HDTV over terrestrial (over the air) networks, cable networks and satellites.

This paper provides a review of possible applications of the exploitation of the 20/30 GHz frequency bands in satellite communication systems. The availability of new higher frequency bands extends to the satellite networks the concept of 'broadband communications', providing new possibilities of integration with the wideband terrestrial networks. Three main applications are here analyzed: (a) the multipoint videoconference; (b) the LAN-to-LAN interconnections; (c) the HDTV service. The work is based on the utilization of the 20/30 GHz ITALSAT satellite, that was designed and developed by Alenia Spazio in the frame of a national program. ITALSAT is in orbit since January '91, providing two different coverages of the Italian territory. In particular, the ITALSAT satellite is considered here as baseline for the system design of the applications analyzed in the paper. The important problem related to the wave propagation atmospheric effects at these frequency bands is also addressed. The heavy fading experimented at Ka-band can have particular impact on system design load. To this purpose, a fading countermeasure technique is selected and analyzed based on an adaptive Up Path Power Control (UPPC) concept.

This paper describes a study for digital transmission system of HDTV signals through the satellite channel. This system has been designed to utilize QPSK modulation via 36 MHz bandwidth Ku band transponder. Convolutional code of rate 3/4 and (255,239) Reed-Solomon code are applied for the purpose of error protection. The system provides the transmission rate of 40 Mbps for video. In order to achieve such video date rate, we studied adaptive noise reduction filter, adaptive motion compensation and adaptive formation of blocks for DCT. This method provides a good picture quality of HDTV signals at the rate of 40 Mbps. The results of our study prove that the system is feasible. At present, we are assembling the codec which performs source coding and part of channel coding in this system. This codec can effectively confirm the performance of the basic algorithm in MPEG2.

A system is described for the transmission and routing of uncompressed digital HDTV signals for use within studio centers. Interfaces have been developed in RACE Project 1081 which allow the transmission of uncompressed digital HDTV signals on optical fiber using the CCITT synchronous digital hierarchy (SDH). Two channels of HDTV or a combination of HDTV and other standard video formats can be carried in a single STM-16 signal. The interfaces can be used with the broadband customer premises network developed in RACE Project 1036 which uses a combination of optical wavelength and electrical time-division multiplexing (WTDM). This network can satisfy all the routing requirements of a modern studio center but also has the capability of routing HDTV signals. The network has a total payload capacity of over 38 Gbits/sec and can therefore carry up to thirty-two HDTV signals. The results of recent experiments are described where the HDTV interfaces were used with a partially populated WTDM network.

Digital HDTV broadcast differs from digital point-to-point transmission in that different receivers have reception condition which depend on the distance from the emitter and on the others disturbances (multipaths, shadowing, etc...). Therefore, a very robust transmission concept taking into account the graceful degradation effect is required. The aim of this article is to show the possibility of designing a transmission scheme supporting three-level hierarchical HDTV signal: HDTV, TV, and Personal Video PV for stationary and portable receivers, respectively. The system is based on a combination of Multiresolution Modulation and Multilevel coding using the principle of OFDM to combat against multipath interferences. The system is designed such that different receiver can receive its desired signal with different complexity according to the image quality. The system also supports graceful degradation.

The video codec presented here will aim at transmitting HIgh Definition Video (HDV) over wide band transmission channels, such as B-ISDN. Primarily, the HDV codec is adapted to the switched broad band network, VBN (Vermitteltes Breitband Netz), of the German Bundespost Telekom with a gross bit rate of 139,264 Mbit/s. To cope with noise effects, appearing in most transmission channels, and especially in satellite links, a robust Forward Error Correction (FEC) unit has been included in the codec.

The transmission of HDTV picture signals is usually carried out by bit rate reduced coding. The compression factor must be kept low for contribution quality. An HDTV codec based on DPCM processing was built to feed high performance video links including the transmission of audio channels. The data rate at the codec output is 648 Mbit/s.

This paper describes an algorithm for frequency scanning (FS) of DCT coefficients which has been submitted to MPEG (Motion Picture Experts Group) in contrast to the well-known block scanning technique. It presents the principle and the coding method of the new scanning technique. With respect to coding the scanned coefficients, an entropy coding method, called modified universal variable length coding (MUVLC), which is derived from the already published UVLC technique, is employed. In the test model (TM) system of MPEG the coding efficiency of both MUVLC and the MPEG VLC was compared. The results showed that the MUVLC technique gives higher coding efficiency and more flexibility in a broadcast environment.

The phasecorrelation algorithm--as a method for motion estimation--is an important component of an HDTV system. The advantage of hardware realization of this algorithm for efficient real time processing--in opposite to blockmatching--is the possibility to process multiple pixels per system clock cycle. A suitable partition for a phasecorrelation chipset is going to be proposed. To minimize the motion estimators physical volume all external components except RAMs, have been integrated on three specialized chips.

The transmission of Digital High Definition Television signal over terrestrial broadcast channel imposes severe requirements on the signal. The system has to handle very large variations of Signal-to-Noise Ratio depending on the distance between transmitter and the receiver and the quality of receiving equipment e.g. antenna. It should provide the HDTV quality picture whenever possible and viewable picture in very poor conditions. One possible solution to this problem is the system providing easily retrievable standard resolution signal embedded in the HDTV signal. In this paper we consider the design of the channel coding part of the system as well as joint optimization of source and channel coding. The results of the simulations of such a system are also included.

This paper describes the experience gained and the planned activities of Italian industries in the HDTV via satellite field. Italy made its experience in digital HDTV transmissions via satellite during the HDTV demonstrated carried out in the occasion of 'Italia '90' and the Olympic Games 'Barcelona '92'. The main topics addressed are the adoption of suitable modulation and coding; the identification of frequency bands for wideband HDTV emission in Europe; the definition of services requirements (quality and availability); the evaluation of technological aspects in the system implementation including all critical elements of the satellite links.

Video compression is achieved by leaving out unnecessary information in an image. This is done by leaving out information that can be predicted (entropy encoding) or by leaving out information the visual system cannot use. Entropy encoding procedure on exact replica of the original image until the image unpredictability exceeds the channel capacity--at which point it fails. Entropy encoding has been discussed widely in the literature. Encoding based on the limits in perception produces an image in which there are measurable differences between the original and reconstructed image. However, these differences are kept below the thresholds of perception. A wealth of information exists in the vision research community on perception thresholds for various visual stimuli. This paper will review these perception limits and show how they have been used or may be used in visually lossless compression.

We describe a new approach to image sequence coding based upon variable-rate entropy- constrained subband coding (ECSBC) and, furthermore, develop the corresponding practical implementation of this ECSBC scheme for fixed-rate channels by extending previously developed adaptive entropy-coded quantization (AEC) techniques. In particular, a buffer- adaptive arithmetic-coded implementation of the ECSBC scheme, denoted ECSBC/AEC, is described which completely eliminates the generally associated encoder buffer overflow/underflow problems, even with a very small encoder buffer. This scheme utilizes hierarchial motion-compensated prediction in a backward-adaptive interframe coding system. Color image sequences are encoded in the YIQ domain. We demonstrate that this ECSBC/AEC scheme operating on real-world image sequences performs very close to the limiting performance achievable only with an encoder buffer of infinite size. Furthermore, we demonstrate that HDTV-quality image sequences can be encoded at bandwidths consistent with existing broadcast television systems. Finally, we show that this scheme delivers extremely robust performance under source mismatch conditions for both video-conferencing and HDTV video material.

This paper describes an advanced method for motion estimation and interpolation for frame rate conversion and slow motion generation of HDTV. The concept is based on a 3-D space- and time-variant filter that is adapted to the motion. Motion is estimated by using a modified block matching algorithm. The main problems of basic motion estimation algorithms are analyzed. The resulting motion information becomes unacceptable, especially in image areas with periodical structures, in the presence of noise or on moving object boundaries. The target of this approach is to increase the reliability of motion information in those image areas and the detection of occluding, appearing and disappearing areas. The spatio/temporal cut-off frequencies of the interpolation filter are controlled taking into account this information. Results of computer simulation are presented on a set of sequences, which are critical with respect to the spatial resolution and the motion complexity.

Packetized video is likely to be one of the most significant users of bandwidth in future high- speed digital networks. In this paper we focus, in a unified manner, on the effects of packet losses, and the resulting error propagation, for a particular entropy-constrained subband coding (ECSBC) scheme employing hierarchical motion-compensated prediction (HMCP). We make use of the associated operational rate-distortion function to assess the quality of transmission that can be sustained under relatively low loss conditions representative of asynchronous transfer mode (ATM) networks. We show that the use of error-correcting codes helps in providing adequate protection for low-to-moderate packet loss. To capture the property of correlated loss in a network, a Markov chain model is used to represent packet losses.

This paper develops some aspects of a structural texture design. We combine continuous and digital geometrical primitives with digital images. A structural textured image is represented as a finite set of texture elements (TEs) and placement rules which govern the spatial relation between them. 1D or 2D geometrical primitives in the Euclidean plane are utilized. The primitive is a domain of a tone function F of 1D and 2D texture elements TEs. We consider the design of different texture pictures: (1) simple structural texture (with one TE), (2) structural textures with nonlinear placement rules, (3) structural textures with few TEs, (4) noised texture, (5) linear textured image. These models allow to synthesize different textured images which have a structural nature.

Block matching algorithm has been widely employed in motion estimation, because of its simplicity of implementation. Among them the full search algorithm gives the optimal solution within the search area except the exhaustive computational cost. On the contrary, other fast search algorithms have the shortcomings of finding the local minimum due to coarse sampling positions. In this paper two issues, speed and local minimum, are dramatically improved by using the line projection and local minima. The proposed algorithm is constructed of two stages. First some candidates, local minima, for global minimum are selected by using the full search on the line projection data. And then these candidates are checked by general block matching algorithm. The performance of the proposed algorithm approaches that of the full search algorithm, but the computational cost is much less than the full search algorithm.

This paper presents an advanced video system which is based on the Alcatel HDTV codec 1741 VC. Broadcast of video signals in a point to multipoint network is upgraded by an interactive control and information interchange between the HDTV terminal side and the HDTV source. ATM cells are used to send the information through the network, advancing future and gaining access to next switching technology. Interface adaptation of HDTV codec to ATM network at 155 Mbit/s is accomplished in such a way that a bidirectional link is supported over the point to multipoint network. Main advantages of this novel system are the sharing of the HDTV coder by multiple decoders and the interactivity added to the terminal end. These features enhance network attractiveness in terms of cost and performance and, finally, gives an extra service of multiple applications.

This paper presents a multimedia communication system supporting group work that exploits the capability of an ATM network system. The system consists of group-work-supporting application software on a workstation and an additional hardware unit that is connected to the external bus of the workstation. This paper begins by presenting the service concept of the system through an analysis of group work and an network properties. Next, it discusses an application software scheme required by the above concept, a multimedia handling that matches its properties, and linkage of group work control and media handling. Finally, it presents the results obtained through implementation of a prototype system which enables multipoint connection through 21 locations and is capable of displaying 11 videos at the same time.

This paper produces problems to solve to develop applications handling super high definition (SHD) images and target values which should be achieved for wide range of SHD image applications from the perspective of the applications and system integration. We classify SHD image applications from the viewpoint of the image characteristics and its interactiveness. For example, medical imaging is one thing electric musium is another. Then problems of technical element that is DBMS, transmission, image compression, etc., about developing applications are clarified.

The efforts so far deployed by European industry engaged in the development of High Definition Television (HDTV) under EUREKA 95 have been mostly based on broadcast applications. These HDTV applications have also led to developments outside the broadcast domain, notable in high definition still imaging. The workstations resulting from these developments might be boosted by accelerator cards in order to be able to process standard or HD television signals. This evolution will require several developments: workstations much more powerful than today's; ASIC and/or multiprocessor developments with the associated parallel programs, and interfaces with other HD equipment, i.e. HD input, genlock and output cards.

Users demands from high-end imaging applications often require picture resolutions beyond the resolution provided by todays High-Definition Television (HDTV). Due to the advances in technology the design of workstations allowing digital processing and realtime visualization of Super High-Definition Television (S-HDTV) image sequences is now possible at affordable expenses. The paper describes the ISP 1500 system under development for storage, processing and realtime visualization of S-HDTV image sequences.

This paper presents a new media integration platform based on super high definition (SHD) digital images and a high performance image processing system that adopts parallel digital processing. The new platform will encourage the integration of all existing media to realize rich and realistic visual communication over B-ISDN. SHD images have a resolution of more than 2048 X 2048 pixels and the frame rate is more than 60 frames/sec. To achieve the real-time compression of SHD moving images, parallel signal processing systems with a peak performance of 0.5 Tera Flops will be necessary. The specification requirements, focusing on the digital signal processing systems needed to achieve SHD image communication, are discussed.

The progress that has been made to date in High Definition Television (HDTV) and related microelectronic technologies in various countries of the world has provided a positive assurance of the success of these technologies. Employing these mature technologies, and without the need of any component development, an imaging product has been conceived which in practical use can provide Aerial Imaging and Surveillance capabilities with unsurpassed performance, resolution, and precision. This product is described and its key parameters are identified. It acquires the surveillance data from an aircraft, and after digitizing and compression passes it on to an image processing center that is located in the ship via a satellite channel. It can provide additional features that the current equipments have difficulties with; these are identification and motion of the undesirable target that can be picked up in the surveillance process.

The general content of the contribution will aim at a Framestore-/Interpolation Structure, being the central component of every Digital Video Effects device--also known as DVE. Special Transformation Mathematics that is necessary for a defined shift of each Picture Element in 3D-Space is well known from graphics literature and will not be discussed in greater detail. More important for the presentation is the very seldomly described way of implementing those algorithms in hardware. A simple example for the still huge amount of necessary hardware will be the description of the structure referenced to a digital image manipulator for the first mixer with integrated DVE working in the proposed european HDTV 1250 Line-Interlaced Standard.

For the purpose of digital recording of HDTV signals (EUREKA standard, 1250/50/2:1) a codec has been realized (HDI-codec) which can reduce the initial data rate from 1,152 GBit/s down to one fifth. According to the desired reduction, the playtime of a digital VCR (Video Cassette Recorder) can be increased from about 40 - 60 minutes up to at least the length of a feature film. This paper describes the hardware realization of the data rate reduction codec. Algorithms utilized for data rate reduction are adaptive intraframe/intrafield discrete cosine transform (DCT), adaptive quantization and variable length encoding (VLC). Interframe editing, multiple copy and shuttle mode are supported by a special codec architecture.

An on-chip memory architecture for video signal processor (VSP) is proposed. This memory structure is a two-level design for the different data locality in video applications. The upper level--Memory A provides enough storage capacity to reduce the impact on the limitation of chip I/O bandwidth, and the lower level--Memory B provides enough data parallelism and flexibility to meet the requirements of multiple reconfigurable pipeline function units in a single VSP chip. The needed memory size is decided by the memory usage analysis for video algorithms and the number of function units. Both levels of memory adopted a dual-port memory scheme to sustain the simultaneous read and write operations. Especially, Memory B uses multiple one-read-one-write memory banks to emulate the real multiport memory. Therefore, one can change the configuration of Memory B to several sets of memories with variable read/write ports by adjusting the bus switches. Then the numbers of read ports and write ports in proposed memory can meet requirement of data flow patterns in different video coding algorithms. We have finished the design of a prototype memory design using 1.2- micrometers SPDM SRAM technology and will fabricated it through TSMC, in Taiwan.

To project block vectors to the new temporal interpolated fields, the known DFD evaluation is unreliable and responsible for the loss of fine details in the sequences. Two improved measures are presented to reduce the degradations. The basic ideas are the detected reliable pixel vector in structured areas should be remained and the temporal vector consistency is checked together with DFD values for the unreliable pixel vector. Initial simulation results have shown that obvious improvements in the scenery of fast motions. No large additional hardware effort is needed.

Any visual communication system is basically composed of image input and output devices, and the resolution capability of each device plays an important role to the visual feature of an image. One of the major problems is the difference of the resolution, not only the spatial resolution (SR, number of pixels), but the amplitude resolution (AR, number of gray levels), between the image devices, which usually causes the image degradation. To preserve the quality of images, a proper conversion of the resolution corresponding to a type of the resolution difference is necessary. A new conversion method of SR and AR has been proposed which considers the impulse response of image devices. It is based on the principle that pixel information includes both spatial and amplitude information, and derived from the impulse response of image devices. The conversion type of higher output SR but lower output AR was previously proposed. In this paper, we consider the conversion type of lower output SR but higher output AR, which has the possibility of resolution trade-off.

This paper describes the characteristics of flat panel displays that use the waveguide principle with liquid crystal (LC) switching. The requirements and previous achievements with respect to the various components of the display are reviewed. Such components include light sensors, modulation methods, planar waveguides, and LC switching methods. The waveguide display approach seems to offer advantages over matrix addressed LC displays, especially for large display area applications. Two novel LC switching methods for efficient deflection of light from the waveguides are described. The other method is based on long-range surface plasmon excitation, and the other on shear addressing.

The DVSR 1000 and the Viewstore VS 500 are multi-standard video sequence processing and storage systems, capable of handling any currently used video standard as well as synthetic formats which might have to be analyzed, particularly those being under consideration for implementation in the future. The system's extremely high throughput rates for the internal system bus and the various video input/output processors ensure its ability to define, program and process any such video formats, even if not being applied at present. Therefore multi- standard image sequence storage systems are the ideal platforms for HDTV research, picture coding and simulation of future visual communication services.

Parallel image processing systems with numerous processing elements (PE), connected in local and regular net, must be fault-tolerant for better reliability and better yield when realized by VLSI technology. The fault-tolerant array processor described in this paper has a regular array of simple programmable PEs and a distributed hardware sub-system for selftesting, selfdiagnostics, and selfreconfiguration.

A matrix form of the Boolean differential temporal (parametric) operators is proposed. The procedures of preliminary binary image processing (logic filtering, finding of contours) are constructed on this base. This presentation of the operators allows to synthesize the algorithms having a mapping into an architecture of systolic arrays.

A system of basic iterative algorithm for signal processing (vector rotation, determination of vector modulus, intervals between points and lines or planes, coordinates of intersection point of lines, etc.) is suggested. The algorithms are intended for hardware realization in the form of LSI- or VLSI-processors and admit obtaining effective parallel calculations by way of conveyer data processing. The hardware realization of the proposed algorithms, which differs from well-known iterative structures by its accuracy of calculations (1 - 2 orders higher), speed (2 - 3 times faster) and extended functional capabilities, is described.

It is reported on the generalized methodology of signal detection in noise which is based on a comparison of statistical parameters of observation sample from region Z of frequency-time noise space where a signal may be present, and observation sample from region Z* of this noise space and it is known a priori about the latter that the signal is absent in this region.

For improving definition of underwater image, a method using special intervallic series of light pulse, and two acts of emitting and receiving are synchronized exactly is presented. The structure of the device which include a special designed emergent-incident optical shutter is also described. Brief calculation show that the back-scattering light can be reduced, and so the definition of underwater image increased effectively. This method has several advantages, which can make it practical for wide use.