Although facsimile systems have existed for more than 25 years, only recently have we seen explosive growth in the use of facsimile, with a corresponding increase in facsimile standards activity. Current standards work concentrates on improving the quaity and speed of image transmission and on increasing functionality. The goal is to benefit from the developing technology-higher resolution scanners and printers, and improved processing, storage and communication technology. Topics now under discussion indude (but by no means are limited to) transmission of color images, improved compression of bilevel images, facsimile routing, facsimile carried by the Integrated Services Digital Network (ISDN), and facsimile applied to teleconferencing. Personal computers can operate as or can be used to add functionality to facsimile terminals. The result is new standards activity (for example, binary file transfer, database storage and retrieval, computer-facsimile communication, and an Applications Programming Interface). Facsimile and facsimile-related standards activity are described with particular emphasis on function and performance. Image compression algorithms are an important factor. Communication protocols are also addressed.

We have developed a polymer dispersed liquid crystal (PDLC) device for operation in the midinfrared region of the electromagnetic spectrum. This device can be incorporated in thermal imaging systems that utilize pyroelectric vidicons for night vision applications. The infrared electro-optic properties of several PDLCs as a function of substrate, droplet size, film thickness, and applied ac voltage are examined using infrared spectroscopy, electro-optic and differential scattering measurements, and static video analysis.

A number of illuminating systems for film copying are based on weil-known dichroic mirrors and electromagnetic slitsvalves. Light from a lamp iiluminator is reflected separately on two mirrors (red and blue) and the third green channel is subtracted. The intensity of the light in each channel is controiled by an electromagnetic valve, and the final color is formed in the exit slit behind a tapered lightguide by mixing of the three principal colors. A new design for an iiluminator that works in a pulse mode and has no mechanical moving parts is introduced. The principal part of the new iiluminator is a cube-formed electro-optic modulator. The new iiluminator operates at a higher frequency of color modulation, with higher energy transmission (higher by 50% than classical iiluminators), and with a greater number of color modes (73 intensity levels for each principal color). Moreover the design of this new iiluminating module is more compact than the one with mirrors, and the new module can be easily connected via computer control.

Typical one-chip color cameras use analog video processing circuits. An improved digital camera architecture has been developed using a dual-slope AID conversion technique and two full-custom CMOS digital video processing integrated circuits, the color filter array (CFA) processor and the ROB postprocessor. The system uses a 768 x 484 active element interline transfer CCD with a new field-staggered 3G color filter pattern and a lenslet overlay, which doubles the sensitivity of the camera. The industrial-quality digital camera design offers improved image quality, reliability, and manufacturability, while meeting aggressive size, power, and cost constraints. The CFA processor digital VLSI chip includes color filter interpolation processing, an optical black clamp, defect correction, white balance, and gain control. The RGB postprocessor digital integrated circuit includes a color correction matrix, gamma correction, two-dimensional edge enhancement, and circuits to control the black balance, lens aperture, and focus.

Morphological granulometries are generated by successively opening a thresholded image by an increasing sequence of structuring elements. The result is a sequence of images, each of which is a subimage of the previous. By counting the number of pixels at each stage of the granulometry, a size distribution is generated that can be employed as a signature of the image. Normalization of the size distribution produces a probability distribution in the usual sense. An adaptation of the method that is appropriate to texture-based segmentation is described. Rather than construct a single size distribution based on the entire image, local size distributions are computed over windows within the image. These local size distributions lead to granulometric moments at pixels within the image, and if the image happens to be partitioned into regions of various texture, the local moments will tend to be homogeneous over any given region. Segmentation results from segmenting images whose gray values are local moments. Especially useful are the means of the local size distributions. Goodness of segmentation depends on the local probability distributions of the granulometricmoment images. Both exact and asymptotic characterizations of these distributions are developed for the mean image of a basic convexity model.

A total and coherent image processing architecture for G3/G4/ISDN facsimiles is proposed that features high-quality multilevelprocessing by means ofcorrelative area scanning and a softwareoriented processing architecture. This image processing LSI controiler includes a resolution converter and error diffusion halftone processing circuits in 4000 gates. A semi-superfine scanning mode is evaluated, which will be adopted as a new CCITT G3 optional mode.

An efficient parallel structure that uses multirate techniques is introduced for the analysis of images by the method of moments. The basic idea of the proposed method is to split the original image into a set of subband signals using a bank of bandpass filters. The subband signals are then low-pass translated by down-sampling resulting in a set ofsubimages oflower dimensions, which are analyzed separately using the method of moments. In image analysis using the method of moments, the major timeconsuming task is the computation of the moments. Moreover, it is hard to reconstruct details of the images from the moments since only moments of higher orders carry the fine detail of an image and they are vulnerable to white noise, such as the quantization noise. With the proposed technique, the computation time is reduced due to the parailel structure, the computational complexity is reduced by using only the dominant subimage in most applications, and the fine detaiofthe image ifnecessary can be provided by the moments of upper-band subimages.

Centroid calculation provides a means of eliminating translation problems, which is useful for automatic target recognition. A neural network implementation of centroid calculation is described that uses a spatialfilter and a Hopfield network to determine the centroid location of an object. Spatial filtering of a segmented window creates a result whose peak value occurs at the centroid of the input data set. A Hopfleld network then finds the location of this peak and hence gives the location of the centroid. Hardware implementations of the networks are described and simulation resuits are provided.

new technique for progressive image transmission (PIT) is presented that uses a self-supervised back-propagation neural network discrete cosine transform. The transmission sequence is determined using a back-propagation neural network (BPNN) feature importance function. Simulation results show that the PlTsystem can be successfully implemented using BPNN. Veiy good intermediate images are obtained at reasonable bit rates.

new method has been developed to measure cell sizes quantitatively from photomicrographs taken during a freeze-thaw experiment on a cryomicroscope. This method uses a region filling and region labeling algorithm to analyze cryomicrographs. Cells can be easily distinguished from noisy pictures and the sizes of cells can be calculated by the method. The results show that this method is adaptable to the analysis of images in a cluttered environment.

With the development of high-definition television (HDTV) systems came the 16:9 (width to height) viewing image aspect ratio. This is compared to the National Television System Committee (NTSC) standard ratio of 4:3 (width to height). This variation in width-toheight aspect ratio has led to the question of which ratio is preferred by the viewing public. The use of a paired-comparison preferencejudgment experiment is described that was designed to determine whether or not significant differences exist in image preference between the two aspect ratios. Observers were asked to choose a preferred image from a set of two (NTSC versus HDTV) of various image sizes over 84 separate trials. Three separate image types were used in the study: a portrait, a landscape, and a still life. The results indicate that image quality perception is a function of image aspect ratio. The HDTV image was preferred for all three image types.

Several possibilities based on the principles of scattering and polarization of light in liquid crystals (LCs) and PLZT ceramics were studied for the replacement of mechanical shutters and attenuators in optical instruments. A scatter-mode LC was selected as the active material, and two modules were developed for fiber optic attenuatorapplications. In the first, the light transmitted through an optical fiber (input fiber, 100I140 rn) is collimated to a beam using a graded-index microlens (0.25 P, 1.8 mm). This beam penetrates two LC films (7 x 12 mm2) and is received by a microlens at the output fiber. The second, for large-diameter fibers (400 rim), employs conventional asphericallenses to optimize the fiber-to-fiber coupling instead of microlenses. Components were built for both 4100- and 4400-p.m fibers. The maximum transmission is up to 70% and maximum contrasts 10,000:1 (632.8 nm), 2000:1 (830 nm), and 700:1 (905 nm). The rise time is typically less than 1 ms and the decay time less than 5 ms, given a temperature above + 20CC. The contrast remains high from 0 to + 50°C. Both components function with