Abstract not available.

Abstract not available.

Abstract not available.

Abstract not available.

Abstract not available.

Abstract not available.

Abstract not available.

Abstract not available.

Abstract not available.

A new approach for fabricating arrays of electron beam columns by stacking silicon wafers with micron accuracy has been developed. This approach combines the precision of semiconductor processing and fiber optic technologies. A (100) silicon wafer is anisotropically etched to create an array of apertures on the top or bottom of the wafer and four orthogonal v- grooves on both surfaces of the wafer. Precision pyrex fibers align and bond the v-grooves on the top of one wafer to the bottom of the next wafer. This procedure is repeated to create thick structures and a stack of six wafers is used to create arrays of scanning electron microscopes (SEMs). This technique is suitable for fabricating 1 - 30 mm long electron optical columns. The optimum size is determined by the desired array size, operating voltage, resolution, field of view, and working distance. The first wafer contains an array of micromachined field emission electron sources. The next three wafers accelerate and focus the electron beams. The last two wafers in the stack have electrodes to deflect each beam and correct for astigmatism. The performance of an SEM improves as its length is reduced and a subcm 2 keV SEM with a field emission source should have approximately 7 nm resolution.

Arrays of sputtered tungsten thermionic emitters with different shapes have been fabricated. A single filament had an emission current of 10 nAmps and a one hour lifetime. The resistance of a filament decreased an order of magnitude with 25 minutes of annealing. After emission the filament appears to have recrystallized in the emitting area. The surface roughness of the recrystallized area strongly depends on the duration of annealing and emitting time. The variation of filament resistance during operation has been investigated. A thermionic emitter has been used as a source for a sub-cm electron optical column. An array of emitters could be used as a high resolution cathode ray tube if the supports are made from a low thermal conductivity material like glassy carbon with thermal conductivity of 10^-3 watt/cm-K¹.

A computer-controlled scan system has been completed for a sub-angstrom resolution STEM (scanning transmission electron microscope). Controlled by a PC-AT, this system is able to control beam scan, unscan, alignment, and correction of aberration. Using this system, the raster size, shape, rotation angle, and raster position may be controlled by subroutines for image generation/transformation. Computer control allows the simultaneous setting of multiple parameters allowing easy operation of the microscope and saving experimenter time.

An automatic computer-controlled system has been completed for conditioning cold field- emission tips in a new high resolution sextupole-corrected scanning transmission electron microscope (STEM). Using this system, a tip may be conditioned optimally and subsequently operated under careful control, extending its lifetime.

We have used AlF₃ as a negative electron beam resist and dry-etch mask for nanoscale device fabrication in both commercial beam writers and scanning (transmission) electron microscopes. Isolated features down to 15 nm have been made as well as square arrays of 20 nm dots-and-spaces.

Atomic force microscopy (AFM) holds great promise for biological research in that it permits: (1) the imaging of membranes and biomolecules of interest with subnanometer resolution in a physiologic environment and (2) the physical manipulation of biomolecules at nanometer scale (nanotechnology). We summarize our recent successful experience with the molecular resolution of hepatic and cardiac gap junctions, porin channels, and organic microcrystals; and with the physical manipulation of gap junction membranes using the AFM cantilever. AFM may revolutionize our approach to the study of structure-activity and perhaps structure- function in many areas of biological research.

We have developed a stand-alone image processing system with an interface to an APL interpreter on a second generation RISC workstation to analyze digital images from our scanning transmission electron microscopes (STEM). The system is UNIX based and utilizes the X-Windows system to display images and to toggle between different environments.

An image processing algorithm was developed to measure ham cross-sectional areas to predicate ham composition. The fat, lean muscle, and skin areas were measured from the cross section of wholesale trimmed hams using image processing techniques. The algorithm was first designed to enhance the ham image using a convolution operation, followed by a gray level normalization operation. Then, a binary image representing ham fat area was obtained after a segmentation operation. Finally the fat, lean muscle, and the skin areas were measured from corresponding binary images. The results indicate that fat and lean can be detected and measured using image processing techniques based on the difference of light reflectance of fat and lean in the visible wavelength of the electromagnetic spectrum.

The object of this paper is to describe an approach to decompose and describe a complex text/graphic into overlapping (either polygonal or curved) meaningful shapes, and reconstruct the original image from the description file. To narrow and optimize the search, some heuristic decision making functions are implemented into the system. In this paper algorithms are designed to automate the process of generation of loops with minimum redundancy from the bit map of the image and identify those loops thus generated if they are simple ones. For the complex loops, decompose them into simpler interpretable shapes or curved segments. Finally, a succinct description file is established for the whole image. Effectiveness of the algorithm has been evaluated through experiments on a large number of graphics taken on overlapped objects. Results show that the algorithm developed is computationally efficient. Once the image is decomposed and the meaningful component parts (either polygonal or curved objects if regular in shape) are identified, the data reduction achieved through this succinct description is extremely high, thus alleviating the matching process. Even for those silhouettes of curved shape, an approach, called concatenated-arc representation, is developed for their description. With this concatenated arc approach, much fewer number of arc segments are needed than needed by line segment approximation. Shapes reconstructed from these description files match closely with the original ones, even for the very complex curves.

An image system using the shadow moire method for the quantitative inspection of highway pavement surface has been developed. Studies have shown that a single moire image is ordinarily not sufficient to yield complete information about the gradient of a set of contour lines pertaining to a particular area of surface. The pavement surface reconstruction algorithm that has been developed to date uses two moire interferograms of any point on the surface to obtain overlapping information to correctly measure elevation. In this paper, a fast Fourier transform (FFT) method is described which generates accurate surface topographical information from a single interferogram. The fast Fourier transform used in conjunction with the shadow moire technique provides the basis for an automated pavement survey system now under development. This new approach is expected to significantly decrease system hardware and processing time requirements for an automated pavement survey system.

The use of ultrasonic imaging to analyze defects and characterize materials is critical in the development of non-destructive testing and non-destructive evaluation (NDT/NDE) tools for manufacturing. To develop better quality control and reliability in the manufacturing environment advanced image processing techniques are useful. For example, through the use of texture filtering on ultrasound images, we have been able to filter characteristic textures from highly textured C-scan images of materials. The materials have highly regular characteristic textures which are of the same resolution and dynamic range as other important features within the image. By applying texture filters and adaptively modifying their filter response, we have examined a family of filters for removing these textures.

The application of machine vision for product inspection and sorting in the vegetable processing industry is a relatively recent development. Machine vision has been applied to inspection and sorting of manufactured products such as printed circuit assemblies and machined parts for over twenty years. The variability of agricultural products requires extensive processing power and the need to do real-time inspection limits, what RS-170 quality video systems can provide. The market size for these inspection and sorting systems does not allow amortizing R&D costs over a large number of units, therefore, maximum use of off-the shelf processing, imaging, and operator interface hardware is required. Also, meeting the time-to-market window for new products with a small, specialized design group does not allow time for in-house high performance hardware design when those designs are available as commodity items. Three of Computer Controlled Machines' machine vision related products are discussed, the Ear Select Machine, the Green Corn Eliminator, and the Post Husker Sorter.

Time-delay and integration (TDI) CCD sensors have been proven to increase the effective sensitivity in linescan imaging applications. This paper describes the design and initial test results of a 6032 element, 32-stage TDI imager for airborne reconnaissance applications. The device is fabricated using a 3-poly 3-phase NMOS process, incorporating buried channel CCDs throughout. It is one-side buttable to produce an array of over 12,000 contiguous elements and is capable of read rates of over 4000 lines per second. For fast readout, the design incorporates dual horizontal CCDs for a total of four outputs in the abutted configuration. The architecture also allows dynamic selection in the number of TDI stages.

In this paper, different factors that affect the quality of a halftone image are examined. We investigated the quality of constant-gray-level halftones, and also of more complex halftone images with various content and degree of detail. For the constant-gray-level halftones, it was examined how their visual appearance changes with variations in their power spectrum characteristics, and for the more complex halftone images we investigated how different quantitative quality metrics can describe the quality of these images. Since the ultimate judges of image quality are human viewers, the success of the quantitative metrics as quality measures for halftones was assessed by comparing their results with the results of a psychovisual test.

A space- and time-efficient algorithm for extracting the interior of a region defined by its boundary is presented. The algorithm uses crack codes rather than Freeman's codes to represent the border. The crack-code representation is topologically well-behaved, whereas the Freeman code treats the foreground and background asymmetrically. We describe the implementation advantages of several appealing properties of the crack-code representation, including uniqueness of path and invariance under complementation. The resulting algorithm is intuitively appealing, efficient, and simple to implement. It has been routinely used as a part of a semi-automatic image segmentation program for MRI image processing.

In this paper, we develop a model-based algorithm for obtaining the maximum a posteriori (MAP) estimate of the displacement vector field (DVF) from two consecutive image frames of an image sequence. To model the DVF, we develop a nonstationary vector field model called the Vector Coupled Gauss-Markov (VCGM) model. The VCGM model is well suited for estimating the DVF, since the resulting estimates preserve the boundaries between the moving and stationary areas in an image sequence. Thus, the model-based MAP estimator that we develop has the ability to adapt to the discontinuities of the DVF. Experiments with a typical video-conferencing sequence demonstrate the superior performance of the proposed algorithm with respect to prediction error and robustness to occlusions.

In this paper, the invertibility of the morphological representation of binary images is investigated. We provide necessary and sufficient conditions for the exact reconstruction of a binary image from its morphological representation. The necessary and sufficient conditions obtained are given by a restriction defined in terms of the morphological thinning transformation. Subsequently, sufficient conditions, for the exact reconstruction of a binary image from its morphological representation, given by a direct restriction on the morphological thinning transformation, are determined. Additionally, we derive necessary conditions, for the exact reconstruction of a binary image from its morphological representation, given by a direct restriction on the morphological thinning transformation.

Contour tracing in image processing is to extract linear features (lines and curves) from images that are useful to intermediate and high-level vision tasks. One difficult problem in contour tracing is to determine the tracing direction at branch points. In this paper a collection of general purpose decision making strategies at branch points is discussed along with a method to combine them into a single criterion that is in fact a weighted combination of those criteria that are considered. Experiments show that results obtained using different branching criteria indeed reflect the need of combining them together, although automatic determination of the weighting factors remains unsolved and needs further study.

Most computer vision algorithms need a good edge description of the scene. The edges are used for high level object recognition. The image of the object is preprocessed to extract edges. Often the extracted edges are short and noisy. This paper describes an algorithm that groups edgelets together to aid in higher level vision object recognition. The input consists of two dimensional and three dimensional range data points from the image of the object. This data is used in testing the edges for grouping properties. These properties include parallelism, collinearity, proximity, and segment length. Short edges are combined into longer line segments using the above criteria. Results are reported for synthetic and real data.

In this paper, we present an edge-based stereo vision algorithm that treats the stereo correspondence problem as a resolution problem rather than a searching problem as in earlier work. The stereopsis is obtained through `fusion' rather than `point-to-point' matching. Rayleigh criterion for resolution diffraction pattern is applied in the fusion process, and the disparity information of the entire image is obtained in one convolution operation. The computation burden is largely reduced. This algorithm uses cross-looking cameras model so that the fixation point can be used as a reference point to register the two images. The edge filter used is developed according to the schematic model of cortical simple cells in the human visual system. During the fusion process, an additional procedure imitating the `optic decussation' in human visual processing is performed so as to minimize the ambiguity due to surface occlusion.

In this paper a new approach based on the subband decomposition (SBD) for compression and progressive transmission of images is presented. The novelty of this approach is that the transform coefficients of all image subbands are coded and transmitted in absolute magnitude order. The ordered-by-magnitude transmission is accomplished by using partition priority coding (PPC) a new source coding method that allows the simultaneous transmission of the coefficients of a data set in an ordered fashion along with position information. Using this approach, coding and transmission are adaptable to the characteristics of each individual image and therefore very efficient. Another advantage of this approach is its high progression effectiveness. Since the largest transform coefficients that capture the most important characteristics of images are coded and transmitted first this method is well suited for progressive image transmission (PIT). Further compression of the image data is achieved by adaptive multiple subsources distribution technique based on arithmetic coding. Experiments are presented where the new SBD approach was tested. It is shown that it compares favorably with previously reported discrete cosine transform (DCT) and subband image codecs and is also very effective for progressive image transmission.

A novel approach for template matching is presented in this paper. This matching scheme is a specialized implementation of a general method for signal expansion by non-orthogonal bases. The basic principle is called expansion matching and involves signal expansion into a set of non-orthogonal template-similar basis functions. It is demonstrated that this matching technique is quite robust in conditions of noise, superposition, and severe occultation. A new and more practical discriminative signal-to-noise ratio for matching is proposed, and it is shown that expansion matching maximizes this ratio. Theoretical and experimental comparisons of expansion matching and the widely used correlation matching, (also known as matched filtering) demonstrate the superiority of our approach. It is also shown that the special case of expansion with a dense set of self-similar basis functions is equivalent to signal restoration. The generalized non-orthogonal expansion as well as the expansion matching can be implemented by our recently developed lattice architectures. Expansion matching can be implemented by restoration techniques as well. The lattice architectures are based on the central limit and can perform multiple-scale Gaussian smoothing of signals in parallel realtime. An adaptive configuration of these lattices is used to expand signals into any arbitrary bases set composed of Gaussian sets arranged in a wavelet configuration.

A brief summary is given of each of the 66 papers presented in the eight sessions at Optical Engineering Midwest 1992 held on March 19, 1992 at the Illinois Institute of Technology in Chicago, Illinois. Each session is also related to some national topical issue.