This paper is a revised version of an article by the same title and author which appeared in the April 1991 issue of Communications of the ACM. For the past few years, a joint ISO/CCITT committee known as JPEG (Joint Photographic Experts Group) has been working to establish the first international compression standard for continuous-tone still images, both grayscale and color. JPEG’s proposed standard aims to be generic, to support a wide variety of applications for continuous-tone images. To meet the differing needs of many applications, the JPEG standard includes two basic compression methods, each with various modes of operation. A DCT-based method is specified for “lossy’ ’ compression, and a predictive method for “lossless’ ’ compression. JPEG features a simple lossy technique known as the Baseline method, a subset of the other DCT-based modes of operation. The Baseline method has been by far the most widely implemented JPEG method to date, and is sufficient in its own right for a large number of applications. This article provides an overview of the JPEG standard, and focuses in detail on the Baseline method. 1

The Discrete Wavelet Transform (DWT) decomposes an image into bands that vary in spatial frequency and orientation. It is widely used for image compression. Measures of the visibility of DWT quantization errors are required to achieve optimal compression. Uniform quantization of a single band of coefficients results in an artifact that we call DWT uniform quantization noise; it is the sum of a lattice of random amplitude basis functions of the corresponding DWT synthesis filter. We measured visual detection thresholds for samples of DWT uniform quantization noise in Y, Cb, and Cr color channels. The spatial frequency of a wavelet is r 2 -l , where r is display visual resolution in pixels/degree, and l is the wavelet level. Thresholds increase rapidly with wavelet spatial frequency. Thresholds also increase from Y to Cr to Cb, and with orientation from low-pass to horizontal/vertical to diagonal. We construct a mathematical model for DWT noise detection thresholds that is a function of level, orientation, and display visual resolution. This allows calculation of a "perceptually lossless" quantization matrix for which all errors are in theory below the visual threshold. The model may also be used as the basis for adaptive quantization schemes.

This paper describes a new technique to implement operations on compressed digital video images that allows many image manipulation operations to be performed 50 to 100 times faster than the corresponding algorithms operating on uncompressed images. This is accomplished by performing the operations directly on the compressed data in the DCT domain. In this paper, we show how to transform image space operations into DCT space operations, we describe several representative algorithms in the DCT domain and report their performance. 1. Introduction Multimedia applications that operate on audio and video data will enable many new applications of computers. For example, a collaborative work system can include video conferencing windows, or a hypermedia training system can include audio and video instructional material. While most research on multimedia applications has focused on compression standards ([3], [9], [10], [14]), synchronization issues ([11], [12]), storage representations ([12...

Texture mapping has become indispensable in image synthesis as an inexpensive source of rich visual detail. Less obvious, but just as useful, is its ability to mask image errors due to inaccuracies in geometry or lighting. This ability can be used to substantially accelerate rendering by eliminating computations when the resulting errors will be perceptually insignificant. Our new method precomputes the masking ability of textures using aspects of the JPEG image compression standard. This extra information is stored as threshold elevation factors in the texture’s mip-map and interpolated at image generation time as part of the normal texture lookup process. Any algorithm which uses error tolerances or visibility thresholds can then take advantage of texture masking. Applications to adaptive shadow testing, irradiance caching, and path tracing are demonstrated. Unlike prior methods, our approach does not require that initial images be computed before masking can be exploited and incurs only negligible runtime computational overhead. Thus, it is much easier to integrate with existing rendering systems for both static and dynamic scenes and yields computational savings even when only small amounts of texture masking are present.

Abstract—The direction-adaptive partitioned block transform (DA-PBT) is proposed to exploit the directional features in color images to improve coding performance. Depending on the directionality in an image block, the transform either selects one of the eight directional modes or falls back to the nondirectional mode equivalent to the conventional 2-D DCT. The selection of a directional mode determines the transform direction that provides directional basis functions, the block partitioning that spatially confines the high-frequency energy, the scanning order that arranges the transform coefficients into a 1-D sequence for efficient entropy coding, and the quantization matrix optimized for visual quality. The DA-PBT can be incorporated into image coding using a rate-distortion optimized framework for direction selection, and can therefore be viewed as a generalization of variable blocksize transforms with the inclusion of directional transforms and nonrectangular partitions. As a block transform, it can naturally be combined with block-based intra or inter prediction to exploit the directionality remaining in the residual. Experimental results show that the proposed DA-PBT outperforms the 2-D DCT by more than 2 dB for test images with directional features. It also greatly reduces the ringing and checkerboard artifacts typically observed around directional features in images. The DA-PBT also consistently outperforms a previously proposed directional DCT. When combined with directional prediction, gains are less than additive, as similar signal properties are exploited by the prediction and the transform. For hybrid video coding, significant gains are shown for intra coding, but not for encoding the residual after accurate motion-compensated prediction. Index Terms—Direction-adaptive entropy coding, direction-adaptive transform, discrete cosine transform, image coding, transform coding. I.

any standard image compression techniques i apply the Discrete Cosine Transform (DCT) to the mage, then quantize the resulting transform coefficients. b For optimal compression, the DCT coefficients should e quantized as coarsely as possible, while allowing . Q minimal visible distortion in the decompressed image uantization of a DCT coefficient induces a noise patr tern over the image consisting of random amplitude eplications of the corresponding basis function. Here r t we measure the detectability of such noise patterns fo hree different size test patterns. Implications of the s w experimental results are discussed. These measurement ill facilitate the design of visually optimized DCT coefficient quantization schemes. 1.

: This paper presents a method for progressive image transmission, which consistently handles the problem of levels of detail and regions of interest. We propose a combined method, that allows the specification of the desired degree of detail for overlapping regions of interest in an image, transmits the necessary image data using a modified zerotree algorithm and supports the refinement of image parts by transmitting only differential data. As an application for small displays, we present the technique of rectangular fisheye views, which takes advantage of the properties of the wavelet representation. Keywords: Progressive Image Transmission, Level of Detail, Region of Interest, Fisheye View 1 INTRODUCTION With the emergence of the World Wide Web, images have become an important means of communicating information in the formerly text-only Internet. Since a substantial fraction of the users connects to the Internet via modems, efficient image transmission methods are crucial for shor...

Inexpensive circuit boards have appeared on the market which transform a normal microcomputer's disk drive into a video disk capable of playing extended video sequences in real time. This technology enables the performance of experiments which were previously impossible, or at least prohibitively expensive. The new technology achieves this capability using special purpose v hardware to compress and decompress individual video frames, which enables a video stream to be transferred over relatively low bandwidth disk interfaces. This paper will describe the use of such devices for visual psychophysics and present the technical issues that must be considered when evaluating individual products.

he term error diffusion has been used in the halftoning literature to describe processes in which pixels' c quantization errors are spread in space to their unquantized neighbors, causing neighboring errors to be negatively orrelated and relatively invisible. The general principle may be extended to the time dimension as well, which we a will refer to as temporal error diffusion. In this paper we will consider the use of temporal error diffusion to meliorate the errors introduced by JPEG image compression of a stream of images. y Keywords: motion, motion JPEG, M-JPEG, error diffusion, temporal contrast sensitivity, visual qualit 1. INTRODUCTION s r The term motion JPEG (or M-JPEG) refers to a video coding scheme in which a sequence of images i epresented as a series of independent still images each of which is coded using JPEG . A number of commercial 1 o i products are available which use this scheme to encode and decode standard video, usually by compressing the tw nterlaced field...

Thrifty scheduling is an algorithm that improves the responsiveness of a stripe-scheduled multimedia server. It increases the determinism of the data-distribution service, reduces the likelihood of high startup delays, and enables an increase in the rated load of the system. A stripe-scheduled media server is a distributed video-on-demand system that load-balances by striping video data across multiple computer nodes and cyclically scheduling the distribution of the data. The server displays highly variable startup delays in response to requests for data streams. These delays are due to clusters of allocated slots in the distribution schedule, which form naturally as the system load increases. Thrifty scheduling is a scalable algorithm that improves responsiveness by allocating streams to schedule slots in a way that reduces the clustering in the schedule. This algorithm has been incorporated into the Tiger video fileserver. Keywords: Video server, video-on-demand, latency reduction, ...