Majid Rabbani

An overview of the JPEG 2000 still image compression standard

Abstract In 1996, the JPEG committee began to investigate possibilities for a new still image compression standard to serve current and future applications. This initiative, which was named JPEG 2000, has resulted in a comprehensive standard (ISO 15444∣ITU-T Recommendation T.800) that is being issued in six parts. Part 1, in the same vein as the JPEG baseline system, is aimed at minimal complexity and maximal interchange and was issued as an International Standard at the end of 2000. Parts 2–6 define extensions to both the compression technology and the file format and are currently in various stages of development. In this paper, a technical description of Part 1 of the JPEG 2000 standard is provided, and the rationale behind the selected technologies is explained. Although the JPEG 2000 standard only specifies the decoder and the codesteam syntax, the discussion will span both encoder and decoder issues to provide a better understanding of the standard in various applications.

Detective quantum efficiency of imaging systems with amplifying and scattering mechanisms

We have analyzed the influence of stochastic amplifying and scattering mechanisms on the transfer of signal and noise through multistage imaging systems in terms of multivariate moment-generating functions. Stochastic amplification of photon noise by one stage of an imaging system is shown to constitute an effective signal to the next, while the underlying photon-noise component is unaffected by a subsequent scattering process. In the case of stationary, photon-limited inputs, these considerations then lead to useful expressions for the noise power spectrum and detective quantum efficiency for multistage image systems. The application of these results to the analysis of radiographic screen-film imaging systems is discussed.

Analysis of signal and noise propagation for several imaging mechanisms

The utility of multivariate moment-generating functions for analyzing the influence of stochastic amplifying and scattering mechanisms on the transfer of signal and noise through multistage imaging systems was shown in a previous study [ J. Opt. Soc. Am. A4, 895 ( 1987)]. Here we extend that study to include cases in which the amplification or scattering parameters are themselves stochastic variables. Each of these cases is illustrated by a physical example drawn from the study of radiographic screen–film imaging. For stationary, photon-limited inputs, expressions for the noise power spectrum and the detective quantum efficiency are derived for each case.

Image Compression Techniques for Medical Diagnostic Imaging Systems

As a result of recent technological advances, a significant (and increasing) number of the images in a typical radiology department are represented in digital form. This includes images that have been captured using digital imaging modalities such as computed tomography and magnetic resonance imaging as well as images that have been digitally converted from film originals as in computed radiography. To provide economical storage of such images and to allow for their efficient transmission over various networks, it becomes necessary to consider digital image compression techniques. In this article, the fundamental concepts of several popular reversible and nonreversible image compression schemes are reviewed. In addition, the performance of the schemes in terms of compression ratio and reconstructed image quality as applied to medical diagnostic images is investigated.

Conditioning contexts for the arithmetic coding of bit planes

The arithmetic coding of the bit planes of an image represented either in the conventional weighted binary form or by the Gray code, is investigated. Two-dimensional contexts based on the pixels from the current bit plane and three-dimensional contexts on pixels from both the current and previous bit planes are studied. Results indicate that for binary bit planes, a three-dimensional template results in a significant bit rate reduction compared to a two-dimensional template of the same size. The savings are much less for Gray encoded bit planes; since the final bit rates are similar to the binary case, the need for Gray encoding is eliminated. The effect of the order of encoding the bit planes are also studied. Comparisons to conventional lossless schemes, such as predictive coding followed by Huffman or arithmetic coding, are presented. >

Bayesian filtering of Poisson noise using local statistics

Images recorded at low-light levels inherently suffer from Poisson noise. A filter based on the maximum a posteriori probability (MAP) criterion is developed to remove this noise. The filter is adaptive; it responds to local changes in image statistics and, thus, removes the noise along the edges without significantly affecting the edge sharpness. It does not require any a priori information about the original image because all the parameters needed for the filter are estimated from the noisy image by assuming local stationarity. Additionally, the simple structure of the filter can be easily implemented in hardware. >

Comparative study of wavelet and discrete cosine transform (DCT) decompositions with equivalent quantization and encoding strategies for medical images

Wavelet-based image compression is receiving significant attention, largely because of its potential for good image quality at low bit rates. In medical applications, low bit rate coding may not be the primary concern, and it is not obvious that wavelet techniques are significantly superior to more established techniques at higher quality levels. In this work we present a straightforward comparison between a wavelet decomposition and the well-known discrete cosine transform decomposition (as used in the JPEG compression standard), using comparable quantization and encoding strategies to isolate fundamental differences between the two methods. Our focus is on the compression of single-frame, monochrome images taken from several common modalities (chest and bone x-rays and mammograms).

Comparison of multiple compression cycle performance for JPEG and JPEG 2000

Ideally, when the same set of compression parameters are used, it is desirable for a compression algorithm to be idempotent to multiple cycles of compression and decompression. However, this condition is generally not satisfied for most images and compression settings of interest. Furthermore, if the image undergoes cropping before recompression, there is a severe degradation in image quality. In this paper we compare the multiple compression cycle performance of JPEG and JPEG2000. The performance is compared for different quantization tables (shaped or flat) and a variety of bit rates, with or without cropping. It is shown that in the absence of clipping errors, it is possible to derive conditions on the quantization tables under which the image is idempotent to repeated compression cycles. Simulation results show that when images have the same mean squared error (MSE) after the first compression cycle, there are situations in which the images compressed with JPEG2000 can degrade more rapidly compared to JPEG in subsequent compression cycles. Also, the multiple compression cycle performance of JPEG2000 depends on the specific choice of wavelet filters. Finally, we observe that in the presence of cropping, JPEG2000 is clearly superior to JPEG. Also, when it is anticipated that the images will be cropped between compression cycles when using JPEG2000, it is recommended that the canvas system be used.

Analysis of image noise due to position errors in laser writers

For raster-written images, the modulated laser exposing beam is scanned across the photosensitive material in a line-by-line configuration. Image noise can be introduced by the writer directly, for example, by the granularity of the photosensitive materials and indirectly, for example, by beam position errors. For analysis of the effect of position errors they must first be related to their resultant exposure fluctuations. Here, position errors are addressed via models of image writing that include several spot/pixel writing schemes. Three types of image noise due to page-scan position error are examined. The effect of low-frequency position errors is described. Exposure fluctuations due to broadband stochastic errors are then addressed. For laser writers using a rotating polygon for beam deflection, the effect of stochastic facet-angle errors is repeated down the image; this results in periodic exposure fluctuations and is the third type of image noise analyzed. Expressions are given for the mean and variance of the exposure error in terms of the statistics of the position error, writing spot profile, and raster sampling distance. The analytical models are then compared with the results of an image simulation calculation. In this way, the exposure error fluctuations are described by their noise power spectra as functions of spatial frequency. After consideration of the sensitometry of the hardcopy recording materials, the exposure errors are then related to the corresponding output density fluctuations.

Learning to produce 3D media from a captured 2D video

Due to the advances in display technologies and commercial success of 3D motion pictures in recent years, there is renewed interest in enabling consumers to create 3D content. While new 3D content can be created using more advanced capture devices (i.e., stereo cameras), most people still own 2D capture devices. Further, enormously large collections of captured media exist only in 2D. We present a system for producing pseudo-stereo images from captured 2D videos. Our system employs a two-phase procedure where the first phase detects “good” pseudo-stereo images frames from a 2D video, which was captured a priori without any constraints on camera motion or content. We use a trained classifier to detect pairs of video frames that are suitable for constructing pseudo-stereo images. In particular, for a given frame at time t, we determine if exists such that It+t̅ and It can form an acceptable pseudo-stereo image. Moreover, even if t̂ is determined, generating a good pseudo-stereo image from 2D captured video frames can be nontrivial since in many videos, professional or amateur, both foreground and background objects may undergo complex motion. Independent foreground motions from different scene objects define different epipolar geometries that cause the conventional method of generating pseudo-stereo images to fail. To address this problem, the second phase of the proposed system further recomposes the frame pairs to ensure consistent 3D perception for objects for such cases. In this phase, final left and right pseudo-stereo images are created by recompositing different regions of the initial frame pairs to ensure a consistent camera geometry. We verify the performance of our method for producing pseudo-stereo media from captured 2D videos in a psychovisual evaluation using both professional movie clips and amateur home videos.