A chaotic output was obtained previously by us, from an Optical Programmable Logic Cell when a feedback is added. Some time delay is given to the feedback in order to obtain the non-linear behavior. The working conditions of such a cell is obtained from a simple diagram with fractal properties. We analyze its properties as well as the influence of time delay on the characteristics of the working diagram. A further study of the chaotic obtained signal is presented.

We demonstrate the operation and rapid reconfiguration of a 12 X 12 Acousto-Optic Photonic Crossbar (AOPC). This AOPC can implement any desired permutation, fan-in, or fan-out interconnection between any subset out of twelve single-mode input fibers into any subset out of twelve single-mode output fibers. The system uses one large-aperture Acousto- Optic Deflector (AOD) driven by a sum-of-tones RF-waveform produced by an arbitrary waveform generator and computed from an experimentally measured lookup table, thus reducing the control complexity of the system. The design, based on the momentum-space technique, includes optical and acoustical rotation for the AOD, in order to optimize the efficiency of the desired interconnections and minimize the undesirable negative first-order acoustooptic Bragg- diffractions. A limitation of this type of systems is the unavoidable reconfiguration (dead) time introduced by the AOD itself, which can result in crosstalk between the individual input channels during that period of time. In this paper, we experimentally investigate the reconfiguration time of this AOPC, by switching between two different crossbar patterns, and then measuring the time during which the detected signal can not be individually resolved for each input channel. Coupling efficiency problems and alignment procedures are also discussed and analyzed.

Spatial Light Modulators (SLMs) are very important components in optical systems. SLMs using arrays of phase modulating pixels are the basis of many new system proposals in adaptive optics, image processing and switching. Ferroelectric Liquid Crystals (FLCs) are especially well suited to devices requiring binary phase gratings and have been widely used. Many of the above applications have been pioneered and demonstrated using `all glass' multiplexed FLC SLMs (128 by 128 pixels) operating in transmission mode. More compact and integrated system will be built using FLC and Silicon SLMs and much work is now focused in this area due to the interest in polarization insensitive FLC binary phase modulation for beamsteering application and arrays of intensity modulator for microdisplays. Here we examine the advantages and disadvantages of a range of different implementations of FLC SLMs for applications requiring arrays of phase modulating pixels. The basis of our comparison is the spatial and temporal distribution of light in the far field diffraction pattern formed by the SLMs. The spatial field is influenced by the flatness and pixel structure of the SLMs, its topology, the surface finish of pixels, the liquid crystal alignment and so forth. Temporal variations arise because of the electronic addressing schemes used to address pixels. Both are important in most applications discussed. The devices considered are all-glass multiplexed FLC SLMs operating in transmissive mode and non- planarized silicon backplane FLC SLM with the mirror flat down on the silicon.

The modified signed-digit (MSD) number system offers parallel arithmetic operations because of its redundancy property. For exploiting the high parallelism of the MSD representation, MSD form and 2's complement binary form have to be converted to and from each other. In this paper, an efficient lookahead-mode conversion algorithm is introduced. The operation can be performed in parallel by binary logic and is thus suitable for optical implementation. The optical system utilizing a single electronic trapping device is suggested to realize the required operations. By programming the illumination of data arrays, any complex logic operations of multiple variables can be performed easily. The main hardware can be stacked to construct a compact optoelectronic processor of versatile functions. A proof-of- the-principle experiment is demonstrated.

A compact two-step modified-signed-digit arithmetic-logic array processor is proposed. When the reference digits are programmed, both addition and subtraction can be performed by the same binary operations regardless of the sign of the input digits. The optical implementation and experimental demonstration using an electron-trapping device are shown. Each digit is encoded by a single pixel, and no polarization is included. Any combinatorial logic can be easily performed without optoelectronic and electro-optic conversions of the intermediate results. The system is compact, general- purpose, simple to align and has a high signal-to-noise ratio.

In this paper, we presented a scheme for complex-valued modular multiplication based on digital partitioning technique. The module (-2)^N-1 is introduced for its simple arithmetic. Both of the operands joining in the complex number multiplication are treated as two N-bit real numbers. Because of its advantage in expressing number, the negabinary number system is adopted. The subtraction in complex number multiplication is converted to addition operation by weighting shifting. Although the direct result takes the looks of a serial of residues, we proposed an algorithm for reconstructing the complex-valued multiplication produce by studying the product decomposing rules. The algorithm can be implemented in parallel manner, we studied the optical implementation based on an incoherent optical correlator and some electronic devices as auxiliary for reconstructing the product. Both the advantage and the disadvantage are analyzed. And the comparison with other optical complex-valued multiplication algorithms is also presented.

Fabrication of high aspect ratio submicron to nanometer range micro structures in LiNbO₃ using a state of the art Schlumberger AMS 3000 focused ion beam system is presented. The submicron structures with about 350 nm width and 1600 nm depth are fabricated by employing XeF₂ gas assisted Gallium ion beam etching with 50 pA of ion beam current. A variety of opto-electronic devices such as micro sensors, directional couplers, extremely compact electro- optic modulators and wavelength filters could be built based on this type of high aspect ratio submicron structures in LiNbO₃, which may lead to the next generation of integrated opto-electronic devices that have higher levels of device integration and enhanced functionality.

We propose star-coupling based on large-core polymer optical fibers. Mixing rods with tapered and untapered ends are employed to form 1 X 7 and 7 X 7 broadcast and mixing couplers. We found that excess loss and coupling uniformity are mixing length dependent and that of a 15 cm length tapered mixing-rod based 1 X 7 coupler can have a 1.72 dB excess loss and a 1.0 dB uniformity. The average excess loss and uniformity for a constructed 18 cm mixing-rod length 7 X 7 star-coupler are found to be 1.38 dB and 1.92 dB, respectively.

Assuming advantages of all-optical 2D switching networks, the question arises which advantages are preserved by the planarization and what are the costs. The planarization of all-optical 2D switching networks show different results depending on (1) the geometry N >= 3 and the (2) implementation by directional couplers and Mach-Zehnder interferometers, respectively. The feed-forward graphs (FFGs), being the complete graph models of planar 2D switching networks, are the starting point for the planarization which is classified into (1) the direct realization where the 2D network is transformed into its isomorphic FFG according to a certain scheme (2) the recurrent realization of 2D networks (a network is represented by smaller networks) and (3) multi-layer realizations. The FFGs are optimized with regard to (4) a minimum number of crossings and (5) minimum skews.

Array illumination is always required for a large-size optical parallel processing system. Talbot array illuminator is highly interesting for a large-size array illumination with high efficiency, excellent uniformity, error-tolerance, and cheap massive fabrication. Theoretically, we derived a new method that can be used to reveal the simple relations between the phase levels of Talbot illuminator and the opening ratio (1/M) of the generated array. Experimentally, we demonstrated an integrated edge-improved Talbot array illuminator. We fabricated one piece of Talbot illuminator with the period of d equals 200 micrometers , opening ratio of 1/2, and binary-phase (0,(pi) ) modulation. One cuboid piece of K9 glass with two-side surfaces coated for high reflection is glued with the Talbot illuminator. The integrated piece is shown to improve the edge uniformity of array illumination at the output plane. Due to wide applications of array illumination, our proposed methods are interesting for miniature of integrated microoptical processing system.

This paper summarized our developments on the integration of miniaturized optical 3D systems using crystal substrates, which include a birefringence-customized stacking technique with the building blocks of double refraction and a photorefractive integration technique with the photorefractive local holograms thermally fixed by CO₂ laser heating. The principles are introduced and the corresponding configurations are given, and the examples of various integrated digital optical computing systems such as logic processor, morphological image processor, and interconnection networks are shown.

After successfully bonding VCSEL arrays to GaAs dummy chips and CMOS chips with three different bonding techniques, the thermal resistance and crosstalk of the bonded VCSEL arrays were measured. The thermal resistance of the VCSELs bonded to a GaAs substrate was found to be as low as 1100 K/W, indicating a high quality contact. Less than 100 K/W thermal crosstalk was also observed in the VCSEL arrays with a pitch of 250 micrometers . The thermal resistance of the VCSEL bonded to a CMOS chip with a standard bonding pad design has also been measured, which is 2490 K/W. The high thermal resistance is due to the dielectric layers underneath the bonding pads.

One-channel representations of hit-miss transform are described. The hit-miss transform can be reduced to be a cross-correlation followed by a thresholding operation after combining the foreground and background structuring elements into one structuring element. A phase-only joint transform correlator is proposed to realize one of the representations. Simulation results are provided.

This paper presents a fast 2D phase retrieval approach used to perform optical phase modulation of a Microelectromechanical Deformable-Mirror (MEM-DM). Traditional solutions to beam splitting, beam steering, and beam shaping (BS³) involve multiple and sometimes costly optical components. For example, beam splitting is normally accomplished with beam splitters, beam steering is normally achieved with gimbaled mechanical devices, and beam shaping is normally done with addressable absorptive devices such as LCDs. Using the phase retrieval algorithm with a desired far-field amplitude pattern as a constraint, a segmented wavefront control device is shown to simultaneously perform the functions of BS³. The MEM-DM used is a foundry micro-fabricated device that is attractive for optical phase modulation applications primarily because of its inherent low cost and low drive voltages. The MEM-DM shapes the beam based on the results of a modified Fienup and Roggemann/Lee phase retrieval algorithm implemented within the system. The optical bench setup and the experimental results for BS³ are presented. Measured experimental data shows good agreement with model simulations. A comparison between analog MEM-DMs and a digitally controlled MEM-DMs will be addressed in this presentation. Overall, experimental results demonstrate the efficacy of the phase retrieval algorithm and one phase control device in solving optics problems normally solved through traditional techniques and multiple devices.

We describe a novel holographic technique for the recognition of 3D objects. We first briefly review an optical heterodyne scanning technique. We will then discuss the use of the technique to create a hologram of the 3D reference object and show how to use this holographic information for 3D object recognition. Finally, we address the important and practical z-invariance of the system and show that the overall scheme of the system is 3D shift- invariant. Computer simulation will be provided for the classification and confirmation of the idea.

A new optoelectronic fuzzy inference system is proposed for processing a large number of fuzzy rules in parallel. The proposed system using spatial light modulator implements various membership functions as well as max-min inference. It has the features of easy implementation and large data processing capability. The membership function decomposition method in the improved fuzzy associative memory is used to save both space bandwidth and accommodate multiple-input fuzzy inference.

This paper explores approaches and issues of developing practical and useful software agents that will communicate and response to their environment. A Multi-AGent architecture that uses SItuation Calculus (MAGSIC) to produce execution plans is designed from scratch. Based on this architecture, a software agent framework that can be reused by agent development is implemented. At present, this framework does not include all the features that are proposed. The framework is tested on Silicon Graphics Inc.'s IRIX Unix machine and PC Linux machine. To achieve automation, a high level language called Resource Programming Language that includes decision constructs, agent communication using KQML performatives, and other features is invented. MAGSIC has some unique characteristics. It is simple in architecture and small in line of code, yet supports a reasoning unit and scales well. An agent implementation like Java Agent Template only supports agent communication but does not support the scripting and reasoning unit. MAGSIC is founded upon a declarative knowledge representation (situation calculus) and clean semantics, equipped with a simple yet powerful approach to procedural attachments (scriptlet). Procedural attachments are tightly, yet flexibly, integrated with the knowledge representation and the reasoning, taking the form of knowledge metadata that annotates the procedural resource. MAGSIC also supporting moving scriptlet and can support mobile agents as well.

Automatic target recognition (ATR) has been a topic of interest for many researchers because of its applications in the fields of defense, manufacturing, health sciences etc. The ability of massive parallelism and high-speed classification of neural network (NN) makes it a good choice for ATR. In this paper, we present a novel ATR approach for targets varying in fine details, rotation and translation using a Learning Vector Quantization (LVQ) NN. The algorithm includes two phases such as the feature extraction and the NN discrimination. The feature extraction algorithm obtains the features of the original and distorted targets in the Fourier-log-polar domain and clusters them into a set of centers. These centers are then applied as inputs to an LVQ NN for training. We explore two distinct discrimination algorithms. In the first algorithm, unrotated target features are applied as training vectors and the network is tested with features of rotated targets. In the second algorithm, the LVQ NN is trained using the rotated images and tested for unknown rotated target features. The algorithm is also applied for the cases of targets varying in fine details and translation and a combination of rotation, fine details and translation.

To construct a hologram numerically, one needs to know the exact depth. In this paper, we present a technique in which the depth information can be extracted from the power spectrum of a hologram. The extraction technique is simulated using mathematical software to confirm the validity of the technique. An example is given.

In crime detection and security verification, fingerprint identification is extremely important. However, in the real- world fingerprint images are not distortion free. In the present work, a distortion-invariant identification method for fingerprint images is developed. The sources of distortion of fingerprint data that are considered are rotation, scaling and erosion. It is shown that rotated fingerprint images with 95% of missing data can still be recognized.

This paper explores a planner that uses knowledgelet and Local Closed World information. Starting with planning theory, basics of planning such as universal base conditional effect of operators, variables in an operator schema, are discussed. The paper ends with a complete problem solver that takes advantage of new features such as knowledgelet, conditional effect of an operator, variables in an operator schema, universal base, and relevance of operators to a problem.

Short-distance optical communication is very much important in numerous applications covering from data-communication (ATM-LAN, Fast Ethernet, etc.) to optical interconnection. Many optical devices are available especially for these applications. The state-of-the-art of optical devices for short-distance, now for premises network are reviewed, and their deployment in different applications are also cited.

We describe a pipelined optical ripple carry adder that adds two words with bits multiplexed by different wavelengths on a single optical fiber. The result of addition is returned to the fiber bus in the same format as the incoming words. The ripple carry arithmetic unit splits the corresponding operand bit pairs off the fiber using WDM demultiplexers. Full adders constructed with semiconductor amplifiers (SOAs) are used to compute the sum and carry from each bit pair and the carry from the next lower significant bit pair. Cross- gain modulation in the SOA is used to: (1) convert incoming wavelengths to a common synchronized wavelength so that destructive interference exclusive ORs may be performed, (2) provide a complement or invert logic operation, (3) perform a NOR logic operation with two or three inputs, and (4) convert wavelength so that the output word is wavelength multiplexed on the outgoing optical fiber in the same format as the incoming words. The sum is computed using a sequence of two exclusive ORs. The carry is computed using NOR gates.

Routing of all-optical 2D switching networks (which connect 2D data whereas the networks span the 3D physical space) is more complex than routing of (planar) 1D networks mainly caused by the (1) combinatorial `explosion' and (2) the arising spatial all-optical >= 3 X 3-switches which are difficult to implement. The routing problem may be subdivided into ((alpha) ) the combinatorial problem of determining the k to generate arbitrary permutations of the inputs at the output of a (rearrangeable nonblocking) network and ((beta) ) the realization of the states of the all-optical >= 3 X 3- switches by the search for waveguide-electrode configurations and voltage adjustments where the paper concentrates on ((alpha) ). The 2D switching networks are (1) projected into plane graphs which--in turn--are (2) mapped onto hypercubes (N equals 4 and 6) and (3) routed by means of the algorithms of (2). Several routing concepts are reviewed and the introduction of several wavelengths is discussed.

We describe a polymer fiber image guided (PFIG)-based optical interconnect circuits which incorporates add/drop capabilities. 2D array data transmitting in segments of PFIG's can be inserted and taken away at various free-space add/drop nodes between these segments. The add/drop nodes are implemented using free-space mini-optical components. A 4-node hybrid circuits using PFIG's to transmit bit-parallel data and free-space components to perform add/drop is experimentally demonstrated. Various power and resolution measurements of the implemented systems are presented.

Optical correlators based on volume holographic associative storage and wavelet matched filtering generally have single input channel. In other words, they can only process one input image at a same time. Based on the fact that a volume holographic correlator is not a shift invariance system, novel volume holographic wavelet correlators with multi- input channels are proposed and constructed in this paper to improve the parallelism of processing. Without adding any component, the method of input plane shifting and the technique of angle multiplexing are combined besides the introduction of wavelet transform. With the correlators, several input images can be recognized simultaneously according to a same system output. Conditions of realizing multi-input channel processing are studied in this paper. The mechanism and prototype of the system are described in detail. The performance of the system in human face recognition is testified by experiment. Promising results are given.

One-step parallel addition and parallel multiplication operations using quaternary signed-digit numbers are proposed with symbolic substitution algorithm using optical cascaded correlation.

A multichannel wavelet correlator is proposed and constructed in this paper, which is based on wavelet transform and mechanism of angle multiplexed volume holographic associative storage in a photorefractive crystal. Problems to enhance parallelism and discrimination of the volume holographic image processing system are studied. The cross-talk noise of the system is analyzed by simulation. Our results show that the cross-talk noise is significantly reduced with the wavelet filtering pre- processing. The correlation outputs can expand from one dimension in a conventional system to two dimensions in our system. Parallelism of the system is greatly enhanced. Wavelet transform has also the function of improving discrimination capability of the system. The conventional correlation is replaced by the correlation between main local features of the images extracted by a same desired wavelet filter, which provides a sharp peak with low sidelobes. Furthermore, other advantages of wavelet transform are introduced. Theoretical analysis and experimental results are given to support our conclusions.

The discriminating ability of the conventional photorefractive correlator is not very satisfactory because the light intensity difference between the correlation beams is not always distinct enough for evaluation. In this paper, we apply the statistical pattern recognition method to the photorefractive correlator to overcome this difficulty. The difference between the method of this paper and the conventional method is that the reference images used to correlate with the input image to be recognized are not the database images, but the eigenimages extracted from the set of training dataset. Since all the cross-correlation results can be collected to form a feature vector for recognition, this method avoids the difficulty mentioned above. In addition, since the number of the eigenimages is much less than the training images, the processing speed can be greatly improved. Furthermore, because the images in the training dataset can be selected to representing some typical distortions, the processor can deal with the distortions to a large extent.

A Multi-grid model using the ADI algorithm (called MADI) is presented to solve the diffusion equation for medical applications and instrumentation development. Nesting multiple blocks of different grid sizes are used in the simulation to save both memory and time. MADI uses the computation results on a coarse grid scheme to provide an approximation of dynamic boundary conditions block-by-block toward the area of interest with finest grid. The precision of MADI is compared with and kept the same as the case of globally uniform finest grid simulation. Algorithm error bound is analyzed by matrix theory. As a conclusion, the thickness of the boundary area and the block width of the finest-grid center block determine the precision. The algorithm is also shown to achieve convergence with only a fixed depth of boundary regions. Finally, the MADI simulation results are compared with both those of the traditional ADI with uniform grids and experimental measurements. The results show the achievement of significant savings (80% - 95%) of memory requirement over uniform-grid computations with essentially no loss of precision. This result demonstrates the feasibility of using such MADI algorithms to achieve high resolution (fine grid, sub-millimeter) computation for realistic organ-size simulations.