A problem which often arises in computer vision is that of matching corresponding points of images. In the case of object recognition, for example, the computer compares new images to templates from a library of known objects. A common way to perform this comparison is to extract feature points from the images and compare these points with the template points. Another common example is the case of motion detection, where feature points of a video image are compared to those of the previous frame. Note that in both of these example, the point correspondence is complicated by the fact that the point sets are not only randomly ordered but have also been distorted by an unknown transformation and having quite different coordinates. In the case of object recognition, there exists a transformation from the object being viewed, to its projection onto the camera's imaging plane, while in the motion detection case, this transformation represents the motion (translation and rotation) of the ofject. If the parameters of the transformation are completely unknow, then all n! permutations must be compared (n : number of feature points). For each permutation, the ensuing transformation is computed using the least-squared projection method. The exponentially large computation required for this is prohibitive. A neural computational method is propopsed to solve these combinatorial problems. This method obtains the best correspondence matching and also finds the associated transform parameters. The method was applied to two dimensional point correspondence and three-to-two dimensional correspondence. Finally, this connectionist approach extends readily to a Boltzmann machine implementation. This implementation is desirable when the transformation is unknown, as it is less sensitive to local minima regardless of initial conditions.

Although the perception of gloss is based on human visual perception, some methods for extracting glossiness, in contrast to human ability, have been proposed involving curved surfaces. Glossiness defined in these methods, however, does not correspond with psychological glossiness perceived by the human eye over the wide range from relatively low gloss to high gloss. In addition, the obtained glossiness in these methods changes remarkably when the curvature radius of the high-gloss object becomes larger than 10mm. In reality, psychological glossiness does not change. These methods, furthermore, are available only for spherical objects. A new method for extracting glossiness is proposed in this study. For the new definition of glossiness, a spatial filter which simulates human retina function is utilized. The light intensity distribution of the curved object is convoluted with the spatial filter. The maximum value Hmax of the convoluted distribution has a high correlation with psychological glossiness Gph. From the relationship between Gph and Hmax, new glossiness Gf is defined. The gloss-extraction equipment consists of a light source, TV camera, an image processor and a personal computer. Cylinders with the curvature radii of 3-30 mm are used as the specimens in addition to spherical balls. In all specimens, a strong correlation, with a correlation coefficient of more than 0.97, has been observed between Gf and Gph over a wide range. New glossiness Gf conforms to Gph even if the curvature radius in more than 10 mm. Based on these findings, it is found that this method for extracting glossiness is useful for the extraction of glossiness of spherical and cylindrical objects over a wide range from relatively low gloss to high gloss.

The demand for large-capacity photonic switching systems will increase as regular broadband ISDN (B-ISDN) spreads and full-motion video terminals replace telephones. Large-scale and economical optical fiber transmission lines have been built based on time-division (TD) multiplexing. To reduce costs, it is important to increase the channel multiplexity of both transmission and switching systems by using TD and wavelength-division (WD) or frequency-division (FD) technologies. We surveyed photonic switching systems' architecture and switching network structures. Switching can be divided into circuit or synchronous transfer mode (STM) switching, and asynchronous transfer mode (ATM) switching. A variety of photonic STM and ATM switching systems based on the two switching technologies have recently been proposed and demonstrated.

The modular and growable photonic ATM switch architecture described in this paper uses both time-division and wavelength-division multiplexing technologies, so the switch capacity can be expanded in both the time and frequency domains. It uses a new implementation of output buffering scheme that overcomes the bottleneck in receiving and storing concurrent ultra fast optical cells. The capacity in one stage of a switch with this architecture can be increased from 32 gigabits per second to several terabits per second in a modular fashion. The proposed switch structure with output channel grouping can greatly reduce the amount of hardware and still guarantee the cell sequence.

Photonic ATM switching systems with Terabit/s throughput are desirable for future broadband ISDN systems. Since electronic LSI-based ATM switching systems are planned to have the throughput of 160Gb/s, a photonic ATM switching system should take the role of the highest layer in a hybrid switching network which includes electronic LSI-based ATM switching systems as its sub-system. This report discusses the state-of-the-art photonic devices needed for a frequency-self-routing ATM photonic switching system with maximum throughput of 5Tb/s. This kind of systems seems to be a moderate system for the first phase photonic switching system with no insuperable obstacle for initiating development, even though none of the devices and technologies required have yet been developed to meet the specifications. On the contrary, for realizing further enlarged throughput as the second-phase photonic switching system, there are huge fundamental research projects still remaining for establishing the technology utilizing the spectrum broadened over 120nm and highly-dense FDM technologies based on homodyne coherent detection, if supposing a simple architecture. "Ultra devices" seem to be the photonic devices based on new tailored materials of which gain and refractive index are designed to realize ultra-wide spectrum utilization.

In this paper, we present a unified model for dialogue understanding involving various sorts of ambiguities, such as lexical, syntactic, semantic, and plan ambiguities. This model is able to estimate and revise the most preferable interpretation of utterances as a dialogue progresses. The model's features successfully capture the dynamic nature of dialogue management. The model consists of two main portions: (1) an extension of first-order logic for maintaining multiple interpretations of ambiguous utterances in a dialogue; (2) a device which estimates and revises the most preferable interpretation from among these multiple interpretations. Since the model is logic-based, it provides a good basis for formulating a rational justification of its current interpretation, which is one of the most desirable aspects in generating helpful responses. These features (contained in our model) are extremely useful for interactive dialogue management.

A new kind of optical local area network (LAN), using a demand assign wavelength division multiple access (DA-WDMA) scheme, has been proposed. The proposed LAN consists of two parts; an ordinary standardized LAN and an overlaid network using wavelength division (WD) channels. The proposed network can provide bit-rate independent communication channels on the ordinary LAN without limiting the capacities for the other channels. It also exhibits upgrade possibilities from present standardized networks. An access controller, which consists of software in addition to the ordinary LAN controller, a digital signal processor (DSP) etc., was developed for DA-WDMA control. The network node operation has been demonstrated using guided-wave acousto-optic (AO) mode converters as a tunable wavelength add-drop multiplexer (ADM).

Among various photonic switching technologies, photonic frequency division multiplexing technology is most promising. In this paper a novel photonic FDM (Frequency Division Multiplexing) system is proposed. The proposed system consists of n (multiplicity of frequencies) independent subnetworks, each of which is identified by a specific frequency, and of which each network topology is identical. When a connection is required by a terminal, the network selects a subnetwork that can afford it, and assigns a frequency representing the selected subnetwork to the terminal. This system eliminates frequency converting devices and traffic concentration equipment, which will reduce the size and cost of the system. A very small sized switching system of very large capacity will be easily realized. In this paper, first we will address the basic concept of the proposed system, and then discuss some technical problems and their solutions concerning network configuration, switch matrix structure, subscriber network configuration, control scheme and frequency multiplicity. Some experimental results are also mentioned.

Student contradictions are the essentials of concepts and knowledge acquisition processes of a student, in the course of tutoring. This paper presents a new perspective to represent student contradictions and a student modeling architecture to capture them. The formulation of a student modeling mechanism enables flexible decision making by using information obtained from students. A nonmonotonic and inductive student model inference system HSMIS has been developed and formulated to cope with modeling contradictions, which basically embodies advanced representation power, sufficiently high adaptability and generality. The HSMIS is evaluated and compared with other representative systems in order to demonstrate its effectiveness.

This paper describes a vision system, which is based on ASIC (Application Specific Integrated Circuit) approaches, for vehicle guidance on highways. After reviewing related work in the field of intelligent vehicles, stereo vision, and ASIC-based approaches, the paper focuses on a stereo vision system developed for intelligent cruise control. The system measures the distance to the vehicle in front using trinocular triangulation. Application specific processor architectures were developed for low mass-production cost, real-time operation, low power consumption, and small physical size. The system was installed in a trunk of a car and evaluated successfully on highways.

This paper describes a noise resistant algorithm for recovering the three-dimensional motion of a rigid object from optical flow. First, it is shown that in the absence of noise three-demensional motion can be obtained exactly by a linear algorithm except in the special case in which the surface of the object is on a general quadratic surface passing through the viewpoint, and the normal vector of the surface at the viewpoint is perpendicular to the translation velocity vector. In the presence of noise, an evaluation function is introduced based on the least squares method. It is shown, however, that the solution which minimizes the evaluation function is not always optimal due to statistical bias. To deal with this problem, a method to eliminate the statistical bias in the evaluation function is proposed for zero mean white noise. Once the statistical bias is eliminated, the solution of the linear algorithm coincides with the correct solution by means of expectation. In this linear algorithm, only the eigenvector corresponding to the zero eigenvalue of a 33 matrix is necessary to find the translational velocity. Once the translational velocity is obtained, the rotational velocity can be computed directly. This method is also shown to be noise resistant by computer simulation.

This paper describes a new algorithm for finding the contours of a moving object in an image sequence. A distinctive feature of this algorithm is its complete bottom-up strategy from image data to a consistent contour description. In our algorithm, an input image sequence is immediately converted to a complete set of quasi logical spatio-temporal measures on each pixel, which provide constraints on varying brightness. Then, candidate regions in which to localize the contour are bounded based on consistent grouping among neighboring measures. This reduces drastically the ambiguity of contour location. Finally, Some mid-level constraints on spatial and temporal smoothness of moving boundaries are invoked, and they are combined with these low-level measures in the candidate regions. This is performed efficiently by the regularization over the restricted trajectory of the moving boundary in the candidate regions. Since any quantity is dimensionless, the results are not affected by varying conditions of camera and objects. We examine the efficiency of this algorithm through several experiments on real NTSC motion pictures with dynamic background and natulal textures.

The character binarization method MTC is developed for enhancing the recognition of characters in general outdoor images. Such recognition is traditionally difficult because of the influence of illumination changes, especially strong shadow, and also changes in character, such as apparent character sizes. One way to overcome such difficulties is to restrict objects to be processed by using strong hypotheses, such as type of object, object orientation and distance. Several systems for automatic license plate reading are being developed using such strong hypotheses. However. their strong assumptions limit their applications and complicate the extension of the systems. The MTC method assumes the most reasonable hypotheses possible for characters: they occupy plane areas, consist of narrow lines, and external shadow is considerably larger than character lines. The first step is to eliminate the effect of local brightness changes by enhancing feature including characters. This is achieved by applying mathematical morphology by using a logarithmic function. The enhanced gray-scale image is then binarized. Accurate binarization is achieved because local thresholds are determined from the edges detected in the image. The MTC method yields stable binary results under illumination changes, and, consequently, ensures high character reading rates. This is confirmed with a large number of images collected under a wide variety of weather conditions. It is also shown experimentally that MTC permits stable recognition rate even if the characters vary in size.

In this paper, we propose a spread spectrum pulse position modulation (SS-PPM) system, and describe its basic performances. In direct sequence spread spectrum (DS/SS) systems, pseudo-noise (PN) matched filters are often used as information demodulation devices. In the PN matched filter demodulation systems, for simple structure and low cost of each receiver, it is desired that each demodulator uses only one PN matched filter, and that signals transmitted from each transmitter are binary. In such systems, on-off keying (SS-OOK), binary-phase-shift keying (SS-BPSK) and differential phase-shift keying (SS-DPSK) have been conventionally used. As one of such systems, we propose the SS-PPM system; the SS-PPM system is divided into the following two systems: 1) the SS-PPM system without sequence inversion keying (SIK) of the spreading code (Without SIK for short); 2) the SS-PPM system with SIK of the spreading code (With SIK for short). As a result, we show that under the same bandwidth and the same code length, the data transmission rate of the SS-PPM system is superior to that of the other conventional SS systems, and that under the same band-width, the same code length and the same data transmission rate, the SS-PPM system is superior to the other conventional SS systems on the following points: 1) Single channel bit error rate (BER) (BER characteristics of the SS-PPM system improve with increasing the number of chip slots of the SS-PPM system, and as the number of chip slots increases, it approaches Shannon's limit); 2) Asynchronous CDMA BER; 3) Frequency utilization efficiency. In addition, we also show that With SIK is superior to Without SIK on these points.

In this paper, we define an attributed random graph, which can be considered as a generalization of conventional ones, to include multiple attributes as well as numeric attribute instead of a single nominal attribute in random vertices and edges. Then we derive the probability equations for an attributed graph to be an outcome graph of the attributed random graph, and the equations for the entropy calculation of the attributed random graph. Finally, we propose the application areas to computer vision and machine learning using these concepts.

This paper describes an adaptive sensing system with tracking and zooming a moving object in the stable environment. Both the close contour matching technique and the effective determination of zoom ratio by fuzzy control are proposed for achieving the sensing system. First, the estimation of object feature parameters, 2-dimensional velocity and size, is based on close contour matching. The correspondence problem is solved with cross-correlation in projections extracted from object contours in the specialized difference images. In the stable environment, these contours matching, capable of eliminating occluded contours or random noises as well as background, works well without heavy-cost optical flow calculation. Next, in order to zoom the tracked object in accordance with the state of its shape or movement practically, fuzzy control is approached first. Three sets of input membership function--the confidence of object shape, the variance of object velocity, and the object size--are evaluated with the simplified implementation. The optimal focal length is achieved of not only desired size but safe tracking in combination with fuzzy rule matrix constituted of membership functions. Experimental results show that the proposed system is robust and valid for numerous kind of moving object in real scene with system period 1.85 sec.

Recently, a family of 4-phase sequences (alphabet {1,j,-1,-j}) was discovered having the same size 2r+1 and period 2r-1 as the family of binary (i.e., {+1, -1}) Gold sequences, but whose maximum nontrivial correlation is smaller by a factor of 2. In addition, the worst-case correlation magnitude remains the same for r odd or even, unlike in the case of Gold sequences. The family is asymptotically optimal with respect to the Welch lower bound on Cmax for complex-valued sequences and the sequences within the family are easily generated using shift registers. This paper aims to provide a more accessible description of these sequences.

A time division duplex (TDD) direct sequence spread spectrum communication (DS-SS) system is proposed for operation in channels with Rayleigh fading characteristics. It is shown that using the TDD method is advantageous because the devices can be designed more simply, the method is more frequency efficient and as a result the systems will be less costly and less power consuming. It is also shown that an efficient power control method can be implemented for the TDD systems. In contrast to the traditional access techniques such as frequency division multiple access (FDMA) and time division multiple access (TDMA) that are mainly frequency limited, the code division multiple access (CDMA) method which uses the DS-SS technique is interference limited. This means that an efficient power control method can increase the capacity of the DS-SS communications system. Computer simulations are used to evaluate the performance of the TDD power control method. Performance improvement of order of 12 to 17dB at bit error rate (BER) of 10-3 can be obtained for different methods of power control. The advantages of the TDD technique for the future DS-SS systems operating in the Industrial, Scientific and Medical (ISM) band are explained in an appendix to this paper.

A new method of multipath diversity combination is proposed for Direct Sequence Spread Spectrum (DS-SS) mobile communications. In this method, the transmitted signal from the base staion is the sum of a number of the same spread signal, each one delayed and scaled according to the delay and the strength of the multipaths of the transmission channel. As a result the received signal at the mobile unit will already be a Rake combination of the multipath signals. This new method is called Pre-Rake diversity combination because the Rake diversity combination process is performed before transmission By this method the size and complexity of the mobile unit can be minimized, and the unit is made as simple as a non-combining single path receiver. A theoretical examination of the Signal to Noise Ratio (SNR) and the Bit Error Rate (BER) results for the traditional Rake and the Pre-Rake combiners as well as computer simulations show that the performance of the Pre-Rake combiner is equivalent to that of the Rake combiner.

In this paper we consider multiuser detection using a neural network in a synchronous code-division multiple-access channel. In a code-division multiple-access channel, a matched filter is widely used as a receiver. However, when the relative powers of the interfering signals are large, i.e. the near-far problem, the performances of the matched filter receiver degrade. Although the optimum receiver for multiuser detection is superior to the matched filter receiver in such situations, the optimum receiver is too complex to be implemented. A simple technique to implement the optimum multiuser detection is required. Recurrent neural networks which consist of a number of simple processing units can rapidly provide a collectively-computed solution. Moreover, the network can seek out a minimum in the energy function. On the other hand, the optimum multiuser detection in a synchronous channel is carried out by the maximization of a likelihood function. In this paper, it is shown that the energy function of the neural network is identical to the likelihood function of the optimum multiuser detection and the neural network can be used to implement the optimum multiuser detection. Performance comparisons among the optimum receiver, the matched filter one and the neural network one are carried out by computer simulations. It is shown that the neural network receiver has a capability to achieve near-optimum performance in several situations and local minimum problems are few serious.