Search

This paper proposes a knowledge model for large-scale knowledge structurization of documents. When an elementary concept such as input for information retrieval is handled as an object, generally a structure composed of the relations in the set of objects is observed. The proposed model generates objects by statistical machine learning of the occurrence patterns of words in the word set and quantifies the relations among the objects by matching to the occurrence patterns. The idea is to approach knowledge structurization in order to achieve systematic understanding by relating pieces of existing knowledge by quantifying the relations among objects as elementary concepts. In this paper, the proposed model is applied to problems in the life science field, where there is a strong requirement for new knowledge with a comprehensive overview, and its application to gene function analysis is discussed. Experimental results on the extraction of disease-related genes as relations between genes and a disease, and on the estimation of gene functions combined with knowledge about organs, are presented and the effectiveness of the approach is demonstrated. © 2004 Wiley Periodicals, Inc. Syst Comp Jpn, 35(8): 21–30, 2004; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/scj.10670 [ABSTRACT FROM AUTHOR]

This paper proposes a Kalman filter based method for high accuracy image restoration. When a Kalman filter is applied to image restoration, the model of the original image affects the accuracy of the restoration. An effective model for restoration depends on the characteristics of the image or the condition of the observed image. On the other hand, the correlation of the original image and the variance of the noise are necessary for image restoration with a Kalman filter. If these parameters are unknown, they must be identified from an observed image which has been contaminated with additive noise. To address the above problems, a method is proposed that identifies the number of pixels used for estimation and their positions. A method to estimate the unknown parameters in the image restoration process is also proposed. In this paper, the performance of the proposed algorithm is verified by simulations. © 1998 Scripta Technica. Syst Comp Jpn, 29(3): 1–9, 1998 [ABSTRACT FROM AUTHOR]

This paper discusses the decision problem concerning security for such cryptographic protocols as the cryptographic key distribution protocols and the digital signature protocols. A cryptographic protocol and its utilization environment are considered secure if an opponent cannot know the secret message or forge a signature successfully by wiretapping or communication function is secure. This paper presents a sufficient condition under which the security problem is decidable. Dolev et el. also investigated the decision problem of the security, but they discussed only the simple case where the cryptographic key is not the object of manipulation (i.e., key generation and distribution are not included in the protocol), and a decision method is presented in which the decision can be made in a polynomial time. By contrast, this paper discusses the problem including the case where the cryptographic key can be an object of manipulation. The key delivery protocol by Ehrsam et al. and the digital signature protocol by Needham et al. satisfy the sufficient condition for the decidability given in this paper. Under the sufficient condition given in this paper, the security can again be decided in a polynomial time. [ABSTRACT FROM AUTHOR]

With the recent widespread application of computer graphics, a management system for graphic data is required which is suited to interactive processing. This paper discusses the management structure for the line data, which can realize a retrieval in an efficient way according to the spatial relations. The dynamic data management structure called the BD time, which was developed by the authors for the multidimensional point data management, is employed. In the proposed system, the BD tree is modified into a form (nonpacket type) suited to high-speed processing, and is extended to suit the line data. In this system, the segment is managed by its centroid and the circumscribed quadrilateral. The BD tree is composed of a set of centroids, where each node has information concerning the circumscribed quadrilaterals for all segments in the subtree. The retrieval is made efficient by the information concerning the circumscribed quadrilateral. In this paper, the data structure for the line data management is discussed first. Then various retrieval algorithms which depend on the spatial relations according to each application are discussed in detail. Finally, the usefulness of the proposed data structure is demonstrated by a retrieval experiment of actual geographical data. [ABSTRACT FROM AUTHOR]

This paper proposes a configuration for a communication control system with a wide range of applications. The proposed configuration is based on the degenerate realization of serial processing (without any restriction of the order) and time-division processing of seven kinds of parallelism included in the communication control processing, as well as on the degenerate realization by shared network bus for the processing request (token) transfer among the components. By the proposed eight kinds of standard configurations, a unified design is made for a wide range of performance from the terminal level to the communication control in a large-scale preprocessing device, The serial processing realized by extracting the protocol-dependent (level) functions and the programmed control provides a flexible means to cope with a wide range of requirements for such functions as odd or delete kinds of links and protocols. This paper outlines the eight kinds of standard configurations and the principle of selection among those together with the derivation of the proposed configurations. [ABSTRACT FROM AUTHOR]

This paper discusses the design of a simple multimicroprocessor system to be built into the dedicated instrument, and also discusses the design for the system. The 8-bit processor sometimes makes it difficult to realize the required function or necessitates a sophisticated technique or experience to realize the function. However, this paper demonstrates that there exist techniques that can easily be incorporated into the processor in the system to realize the required function and performance. The hardware discussed in this paper is flexible so that the result of functional analysis can be realized directly as the structure. The problems of parallel processing that will inevitably arise in this kind of approach are then solved by queue and synchronized registers. Hence, the hardware structure (number and connections of processors and memories) is determined according to the result of structured programming, and parallel processing programs are designed to be performed along the lines of the past single-stream system. The performance of the system can be modified by modifying the number of processors and the hardware structure without completely changing the program, only by reassignment of the memory. From the practical viewpoint, 80- type CPU (Intel 8085A), most widely used at present, is adopted. The result of experimental set-up and simple application examples are shown. [ABSTRACT FROM AUTHOR]

This paper discusses the structural analysis algorithm for line patterns, based on the distance transformation of a line pattern by iterative local parallel processing. Although parallel processors have been applied recently to image processing, the processing of line patterns in the past was based primarily on the transformation of the object pattern into the list structure or symbol strings. Few studies have been made on the parallel processing algorithm for a line pattern. From such a viewpoint, this paper presents parallel-type algorithms for loop counting, loop extraction and arc decomposition, among the procedures for processing line patterns. Properties of these algorithms also are discussed. The method proposed here is based on the distance transformation (Min. type) for line patterns we developed, and can be performed by iterative local parallel processing, in the same way as the distance transformation itself. [ABSTRACT FROM AUTHOR]

Flexibility, extensibility and reliability of a computer network are efficiently achieved by a design with hierarchical architecture. In an actual network system the network design should be implemented by modules corresponding to the layers of the hierarchical architecture in order to real- ize Its inherent advantages. Generally, in highly modularized software the problem of overhead in transferring data between modules has to be solved. This paper presents a method of network software design which is easily implemented by a highly modularized software structure and describes implementation of the architecture through separate communication control and data processing in an operating system. These presentations are made by introducing the methods employed in constructing a local computer network called the S & T net. The methods presented in this paper can be applied generally to the design and realization of network software. [ABSTRACT FROM AUTHOR]

We propose a conversation method called seamless interaction where the viewer can query a CG generated performer at any time while viewing video content based on CG, and can interrupt video content for direct conversation with the performer. In this paper, we explain the important requirements for achieving seamless interaction, and show an example mounting these requirements. Then, we use a subjective evaluation test to compare seamless interaction with images of interaction with CG characters using dedicated conversational CG characters. From these results, we show that seamless interaction is an effective method for viewers when engaged in passive viewing, such as viewing a television program, in terms of ease of interaction and ease of familiarity. © 2007 Wiley Periodicals, Inc. Syst Comp Jpn, 38(13): 1–14, 2007; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/scj.20833 [ABSTRACT FROM AUTHOR]

The trie method is widely used as a dictionary retrieval technique in natural language processing, and the double array method is extremely good in terms of high retrieval speed and compactness as a data structure for implementing the trie method. However, although placing the double array in main storage is effective for fast retrieval, the size of the double array increases as the number of storage keys increases. Therefore, in this paper, the authors propose a divided double array method that can execute retrieval efficiently even when using a small main storage area and can also further reduce the amount of storage by dividing the double array and storing it in secondary storage. However, a problem with this technique is that the number of accesses to secondary storage increases because the divided tries must be scanned. To deal with this problem, the authors proposed a technique that reduces the number of times the divided tries are scanned to maintain the fast access speed of the double array. They performed experiments for sets containing from 70,000 to 230,000 keys, and from the results, it was apparent that the number of accesses to secondary storage during retrieval for a single key was less than or equal to 2, and the storage capacity required for the divided tries was reduced by approximately 50% compared with the undivided trie. © 2004 Wiley Periodicals, Inc. Syst Comp Jpn, 35(8): 72–81, 2004; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/scj.10481 [ABSTRACT FROM AUTHOR]

In this paper, we designed and implemented an intermodule connection architecture, which is required for the development of modules for a multifunctional robot. The proposed and implemented architecture is a dual bus architecture characterized by the fact that a composite blackboard and publish/subscribe service for realizing flat relationships among modules and the dynamic switching of information flows by the modules themselves can be compatible with tight integration and a guarantee of scalability. Through an implementation of conversational tasks, which used speech, acoustic, and image information in an integrated manner, it is confirmed that a tightly integrated system can be developed in a short period. © 2004 Wiley Periodicals, Inc. Syst Comp Jpn, 35(8): 54–65, 2004; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/scj.10585 [ABSTRACT FROM AUTHOR]

In order to create efficient object-based video encoding, the authors have proposed a scene-adaptive region segmentation method that can work with a variety of different types of scenes. In their method, encoding efficiency is improved by adaptively selecting for each frame the optimal method from among several different region segmentation methods determined using Bayesian decision-making based on the video features of a scene. In this paper, the authors describe the Bayesian decision and training methods, and then formalize the image features and the different region segmentation methods. Finally, the authors demonstrate the validity of the proposed approach through experiments. © 2004 Wiley Periodicals, Inc. Syst Comp Jpn, 35(8): 31–44, 2004; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/scj.10593 [ABSTRACT FROM AUTHOR]

This paper proposes a new neural network (the concepts-of-sentences associative neural network, CSANN) which performs memory and recall of sentences represented by a semantic network, as an approach toward natural language processing. The semantic network is represented in terms of the triples representation of concepts. The triples representation of concepts consists of concepts linked by relations in sentences represented by natural language. CSANN handles sentence data, which are difficult to handle since the data size is not unique, by decomposing the data into triples representations of concepts and integrating them into the semantic network format. It is verified by computer simulation that the proposed network can store sentences and recall sentences even if the input information is incomplete. Thus, the effectiveness of the proposed network in natural language processing is demonstrated. © 2003 Wiley Periodicals, Inc. Syst Comp Jpn, 34(11): 43–54, 2003; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/scj.10476 [ABSTRACT FROM AUTHOR]

When dealing with high-dimensional data, it is proper to construct an appropriate feature space consisting of relevant features extracted from given data. Mixture of Principal Component Analyzers (MPCA) is a feature extracting tool in which clustering in the data space and principal component analysis in each cluster are simultaneously done. A probabilistic model of MPCA and its maximum likelihood (ML) inference were presented by Tipping and Bishop (1999). Although the probabilistic formulation can be satisfactorily advantageous within Bayesian inference instead of ML inference, Bayesian inference often requires difficult integrations. Such difficulty comes to be overcome by the variational Bayes (VB) approximation that has recently been developed. In this paper, we present a VB inference algorithm for MPCA, and its application to a recognition problem of handwritten digit images. © 2003 Wiley Periodicals, Inc. Syst Comp Jpn, 34(11): 55–66, 2003; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/scj.10394 [ABSTRACT FROM AUTHOR]

The specification for a communication protocol is generally represented by a finite-state machine, and the operation of the machine is represented by transitions among states. The device implementing such specification is called the implementation under test (IUT). It is important for the IUT that its operation should be verified for all state transitions on the finite-state machine given in the specification. This test is called the conformance testing. In such testing it is important to generate efficiently a route containing all state transitions from the initial state. This problem is called the test sequence generation problem. This paper considers the test sequence generation problem for a communication protocol and proposes an application of an algorithm for satisfiability problem (SAT) that can flexibly handle various constraints, such as the order constraint and the time constraint among multiple constraints. The proposed method is applied to the dynamic host configuration protocol (DHCP) and its effectiveness is demonstrated. © 2003 Wiley Periodicals, Inc. Syst Comp Jpn, 34(11): 20–29, 2003; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/scj.10482 [ABSTRACT FROM AUTHOR]

The paper proposes an improvement of vector search performance for a vector quantization method based on the adaptive orthogonalized transform (AOT). The principle of the improvement is reducing the number of vector comparison by limiting the range of the dictionary. To limit the range of the dictionary, the dictionary vectors are sorted by the angle to the standard vectors in advance. The optimal vector position is estimated by the angle to the standard vectors in accordance with the position tables, and comparison is made around the estimated position. The effectiveness of the proposal is confirmed by numerical experiments. © 2003 Wiley Periodicals, Inc. Syst Comp Jpn, 34(11): 10–19, 2003; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/scj.10491 [ABSTRACT FROM AUTHOR]

This paper proposes a Compton CT technique which accurately reconstructs the image of the radiation source distribution in the object cross section from the data observed by Compton scattering. The source space tree algorithm (SSTA) is proposed as an image reconstruction method in Compton CT. The method consists of reconstructing the source distribution by summing the weights assigned to the voxels intersecting the surface of a cone determined by the Compton scattering angle. In this process, it is necessary to make a decision about the intersection between the voxels composing the back-projection space and the cone surface. In the conventional method, the spherical region is used in making the decision. This may produce an incorrect judgment of intersection, resulting in an artifact on the image. In the proposed method, a cube is used instead of a sphere. This makes the intersection decision more accurate, resulting in reconstruction of the image without artifacts. The effectiveness of the proposed method is verified by simulations. © 2003 Wiley Periodicals, Inc. Syst Comp Jpn, 34(11): 1–9, 2003; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/scj.10496 [ABSTRACT FROM AUTHOR]

This paper considers the energy demand prediction problem and formulates it in the framework of hierarchical Bayesian nonlinear time series prediction with a neural net. As a practical problem, the method is applied to thermal load prediction of a formal energy storage system and the results are presented. © 2003 Wiley Periodicals, Inc. Syst Comp Jpn, 34(11): 101–110, 2003; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/scj.1217 [ABSTRACT FROM AUTHOR]

This paper introduces an application of iris recognition technology, using the iris pattern for horse identification. There are several problems to be solved in horse iris recognition: (1) It is very difficult for horses to remain motionless, which leads to mislocation and loss of focus during image acquisition, so that the images often have poor quality. (2) Pupil size varies significantly between the conditions of dilation (mydriasis) and contraction (miosis). (3) Horse iris patterns are not clear. As a solution for issue (1), we used the reflection of the illumination sources employed for image acquisition and chose adequate images suitable for recognition. For issues (2) and (3), we propose region extraction appropriate to the equine eye structure, a stable coordinate model for pupil variation, and recognition using orthogonal wrinkles in the iris pattern. Recognition experiments based on 100 sets of horse iris data show that highly accurate horse identification is possible. © 2001 Scripta Technica, Syst Comp Jpn, 32(14): 12–23, 2001 [ABSTRACT FROM AUTHOR]

This paper presents a method in which moving obstacles such as pedestrians, bicycles, and automobiles are detected from a moving vehicle using images, and the locations of the obstacles are calculated stably in relation to the vehicle. In the proposed method, stereoptic images are projected on the estimated road plane, and a free-space map is constructed without attempting matching between the stereoptic images. Then, the vanishing point of the motion (FOE) is estimated from the optical flow, and the moving obstacle area is extracted from the image based on the residual error. By projecting the moving obstacle area onto the free-space map, the location of the moving obstacle relative to the moving vehicle is determined. In this method, neither tracking between frames nor matching search between stereoptic images is needed, and the location can be determined stably, even for objects with rapidly varying size in the image, or nonrigid objects with continuously changing shape. An experiment was performed in which the locations of moving obstacles such as pedestrians, bicycles, and automobiles were determined from the input image sequence obtained from a camera on a moving vehicle. The effectiveness of the proposed method was demonstrated. © 2001 Scripta Technica, Syst Comp Jpn, 32(14): 67–76, 2001 [ABSTRACT FROM AUTHOR]

The authors have developed a communications architecture called RHiNET that is capable of high-performance parallel processing by interconnecting PCs and WSs. RHiNET uses not only computers concentrated in one place but also computers used for daily business within buildings and within floors for parallel processing. Networks used by RHiNET should have high effective bandwidths and small latencies for high-efficiency computing, as well as link lengths for connecting the computers within buildings and floors and a free topology. The former requirements are not satisfied sufficiently by the conventional LANs, while the latter is not satisfied by SANs which are used in cluster systems. RHiNET satisfies both conditions by using a switch having a large memory capacity. In this paper, the architecture and implementation of the RHiNET/SW1 switch used by RHiNET networks are discussed. A scheme called virtual cache is used in RHiNET/SW1 such that a large-capacity packet buffer is installed by SRAM outside the switch ASIC. In addition, the performance of the switch is evaluated by a simulator. © 2001 Scripta Technica, Syst Comp Jpn, 32(14): 24–33, 2001 [ABSTRACT FROM AUTHOR]

In general, systems are comprised of a large number of elements and might be very symmetric. In verification of such systems, they are reduced while upkeeping their global structure and verified with the intuitive expectation that the verification of such reduced systems will yield almost the same results as those of the original ones. However, such verifications do not guarantee the success of original system verifications. In this paper, the target is verification of arbitrary scale systems with star-shape configuration which can be expressed by Petri nets. The proposed algorithm explores the reachable state space with no dependency on system scale. © 2001 Scripta Technica, Syst Comp Jpn, 32(14): 41–53, 2001 [ABSTRACT FROM AUTHOR]

We propose in this paper a time-dependent link scheme for hypermedia. The management of the links accompanying updates in hypermedia is a major subject; in particular, it is an important problem in hypermedia dealing with content which depends on time. This scheme will realize the dynamic creation and expiration functions of links as well as the automatic link control function. We have packaged a prototype system called Mille-feuille using the proposed time-dependent link scheme and have confirmed the scheme's validity. This scheme can also be realized by using the extensible markup language (XML). © 2001 Scripta Technica, Syst Comp Jpn, 32(14): 54–66, 2001 [ABSTRACT FROM AUTHOR]

This paper proposes a new motion detection method for image-sequence processing. It consists of hierarchical variable-block-size motion detection with a constant average block size. It makes use of hierarchical images to detect motion vectors. The block size is decided by termination of motion search in a hierarchy level with a small matching error. The termination threshold is decided by the detection result in the highest hierarchy level. It can be set up according to a desired average block size per frame. Efficacy of the proposed method is confirmed by simulations on video coding. [ABSTRACT FROM AUTHOR]

This paper describes genetic algorithms (GAs) applied to Bayesian image restoration. Crossover operation is discussed as a method used to find new search points in GAs. A suitable genotype and an efficient crossover method for a Bayesian image restoration will be proposed. Due to the independency of GAs from objective function forms, it is easy to introduce ideas into the objective functions. The idea of line process proposed by S. Geman and D. Geman [12] is introduced into the objective function, which provides better restored images. As a comparison, a stochastic relaxation (SR) method is applied to the image restoration problem, and the convergence properties of SR and GAs are also discussed. [ABSTRACT FROM AUTHOR]

This paper proposes a quasi-competitive network in which the mass reaction intensity of the elements in the intermediate layer is kept constant as a model that approximates the continuous nonlinear function. As the loss function, the local loss function is considered. It is composed of the error and the local model loss term reflecting the local model size. The learning algorithm of the quasi-competitive network including the change of the number of elements is derived from the local loss function. It is shown by numerical experiment that the quasi-competitive network can form quickly the structure reflecting the target function and has a high generalization power. It is shown also that the high generalization power is due to the constant mass reaction intensity of the elements in the intermediate layer of the quasi-competitive network. [ABSTRACT FROM AUTHOR]

In the neural network for pattern recognition, when the selectivity of the feature-extracting cell is lowered to enhance the generalizing power, a tendency is produced that patterns with similar shapes but belonging to different categories are confused. Because of this property, it has been difficult to en-hance the discriminating power to separate patterns with similar shapes without deteriorating the generalizing power. If patterns are observed, however, focusing on the features which differ between similar patterns, it will be possible to discriminate similar patterns. In the earlier neocognitron, the feature-extracting cell has uniform sensitivity within the receptive field. Then, it is impossible to discriminate patterns by emphasizing the local feature, which is located at a particular position in the receptive field. This paper proposes a method of learning, where the sensitivity of the feature-extracting cell is made variable depending on the position in the receptive field, and the sensitivity is adjusted by the error backpropagation. After the learning, the feature-extracting cell exhibits a high sensitivity at the position where similar patterns have different features. Thus it is made possible to classify similar patterns without deteriorating the generalizing power. [ABSTRACT FROM AUTHOR]

The use of discriminant analysis for controlling software quality using complexity metrics is explored. One of the main problems in discriminant analysis involves selecting the principal components among complexity metrics that affect the classification of software modules into several groups based on quality. This paper focuses on the use of Akaike information criterion (AIC) procedures for selecting these principal components. These procedures are based on the residual sum of squares obtained from the maximum likelihood estimation and the number of complexity metrics. The validity of using AIC procedures is confirmed by Wilks &Lamda; -criterion procedures, taking the discriminant efficiencies (i.e., F-test values) of models and those of complexity measures into account. Experiments show that AIC procedures can identify the model that has the highest discriminant efficiency with a low misclassification ratio and that the maximum F-test value model is an approximate solution of the minimum AIC estimated model. [ABSTRACT FROM AUTHOR]

An understanding of state transitions is essential in real-time software design. However, satisfactory research has not been performed on methodologies to list the necessary states. A method is needed to systematically identify necessary states in order to meet the given requirements. The two points that must be clarified when identifying necessary states are which states are necessary states and what causes lead to considering a state necessary. We begin this paper by clarifying the necessary states in software design that must be introduced into the software when elements other than state parameters external to the software arc required to determine the correct response when multiple responses are obtained from the same input. Next, by focusing on the relationship between the input and the output, we illustrate that the software requirements have a hierarchical structure and selections from alternative realization methods based on the structure cause consideration on states. This requirement structure designates the logical structure of the requirements specified in the design process. Design and specification (implementation) of the requirements make is possible to list necessary states systematically. [ABSTRACT FROM AUTHOR]

In the distributed processing applications, the dynamic load balancing is considered to be important. In most of the past studies concerning the dynamic load balancing, where the load is distributed based on the information during the job execution by the system, the load of a computer is defined as the number of jobs in execution on the computer. In other words, the computer resources are given little consideration. Even if the computer resources are considered, the model is defined as a homogeneous system composed of computers of the same performance, which seems to deviate from reality. From such a viewpoint, CPU and the input/output units such as disk are considered as the computer resources, and are represented by the load vector in this paper. A relatively simple dynamic load balancing algorithm using the load vector is proposed. The method is extended so that it is applicable to the non-homogeneous system composed of computers with the same function bat with different speeds. The proposed method is simulated by a model in which computers are connected through a bus-type network. As a result, various effects were recognized, such as the decrease of the mean response time, the decrease of the variation of the response time, and the effective use of a high-speed computer in the heterogeneous system. [ABSTRACT FROM AUTHOR]

This paper describes algorithms of magnifying/reducing, rotation, and border following for binary document images by using a segmented block code. Document images such as facsimile images are composed of a large number of binary data. Thus it is difficult for a small-sized conventional computer to process efficiently a whole document image in a pixel-by-pixel manner. Our algorithms described herein use segmented block coded data instead of pixel data. In the segmented block coding scheme, blocks which are two-dimensional contiguous black pixel areas are extracted and coded by their locations and shapes. Since this coding scheme compresses the amount of data required to process and produce the data form which is suitable to deal with by conventional computer, our algorithms using this coded data work efficiently. Also, there is another advantage in that the results of magnifying/reducing or rotation of Images are obtained in the form of a segmented block code. This means that we can manage document images in a unified manner. [ABSTRACT FROM AUTHOR]

Information processing in living organisms is carried out by the nervous system. From engineering viewpoint, the nervous system has various interesting features. Study of the hardware neuron model is necessary for its engineering applications. A hardware neuron model by lambda-shaped transistors has bean reported by the authors. This model has a voltage-current characteristic similar to that of living neurons, suitable for engineering applications. However, this model simulates generation of only single action potential in response to constant current stimulus. It is not a sufficient hardware neuron model for transmitting information by pulse rate. This paper describes a pulse-type hardware neuron modal using a lambda-shaped transistor and its applications In information processing. First, analysis of the early circuit leads to a model generating continuous action potential by constant current stimulus. Then it is explained that a hardware neuron model showing ON-response and a Sequence Detector can be constructed with application to temporal feature detection. [ABSTRACT FROM AUTHOR]

This paper discusses the hierarchical properties of the k-neighborhood template [This symbol cannot be presented in ASCII format]-type two-dimensional bounded cellular accepter (abbreviated as [This symbol cannot be presented in ASCII format]-2gCA(k)) on three-dimensional tapes we proposed previously. An [This symbol cannot be presented in ASCII format]-2BCA(k) is s three-dimensional automaton which consists of a pair of a converter and a configuration-reader. The former converts the given three-dimensional tape to the two-dimensional configuration and the latter determines the acceptance or nonacceptance of given three-dimensional tape whether or not the derived two-dimensional configuration is accepted. We investigate how the difference of the neighborhood template of the converter or the configuration-reader affects its accepting power. As a result, it is shown that when the converter Is deterministic, the greater the number of neighbor cells of the converter and the more powerful the configuration-reader, the sore powerful the accepting power of the [This symbol cannot be presented in ASCII format]-2BCA(k). However, this is not always true for the nondeterministic case. As far as this case is concerned, the following results are obtained: (1) Nine-neighbor is as powerful as five-neighbor whatever the configuration-reader combined; (2) there exist classes of two-dimensional automata such that indeed their accepting powers are different, but those of [This symbol cannot be presented in ASCII format]-2BCA(k) using these are identical. [ABSTRACT FROM AUTHOR]

In the videotex communication service system, a new matrix function is planned. The drawing pad may be used in writing the mail, where handwritten patterns can be accepted. However, this type of mall is composed of geometric information, and the conversion to pattern information (G/P conversion) is required in the communication device to produce the output processing in the communication device in the low-level terminal. This paper proposes a G/P conversion method (successive conversion), which suffers less delay in the start of the display and preserves the order of writing. In the successive conversion, the expansion of a geometric primitive and the transmission of the result of expansion are executed alternately. Only the rectangular region containing the part for which the expansion has been made is sent as the pattern data. The proposed method is applied to the drawing mail. It is shown that the processing required in the conversion is twice that of the straight system, where the result is sent after expanding all geometric primitives. It is also shown that the adequate speed of display for the character is 2.7 character/s, if it is displayed following the writing order. [ABSTRACT FROM AUTHOR]

Optical flow analysis is a powerful means to extract an object from the dynamic image. Not only can it extract an object, but it can also recover the three-dimensional structure. However, in practice, there are many cases where the optical flow is difficult to obtain or the motion of the object does not permit a unique interpretation. Based on the gradient method, this paper proposes a method of three-dimensional motion analysis for multiple moving objects which may include an object for which a unique interpretation of the motion is difficult. By the gradient method, the three-dimensional motion par-meters of a rigid object can be estimated without using the optical flow as a least-square-error solution for the system of linear equations, if the three-dimensional structure of the object is already given. Based on the residual square-sum error in the estimation, the image plane is segmented if it contains regions of different motions. If the motions are recognized as being similar, the regions are merged. Thus, the object is extracted by iterating the segmentation and merge on the image plane. The uniqueness of the motion is evaluated by the condition number for the coefficient matrix of the system equations. If the motion of the extracted object can be interpreted uniquely, the displacement vector can be determined accurately by iterating the estimation of the object motion. [ABSTRACT FROM AUTHOR]

As a model to describe the concurrent system with a synchronization mechanism, Holenderski has proposed the synchronized shuffle expression (SSE). SSE is a shuffle expression obtained by adding a synchronized shuffle operator with its closure and projection operator. This paper considers various subclasses of SSE. It is shown that the class of languages represented by the subclass excluding the projection operator is the same as the class of recursively enumerable languages. The hierarchical relations among the subclasses are described. [ABSTRACT FROM AUTHOR]

In the search for correspondence in the stereo vision, a genoral method is to use dynamic programming and search the correspondence for a pair of left and right scanning lines. In this case, the edge is often used as the unit for the matching, but the method has a problem in the processing of occlusion and noise. To solve this problem, this paper proposes a method in which the interval between two edges is considered as the unit of matching. Three components are proposed as the coat evaluation in the interval matching, which are compared by using synthesized images. As a result, it is seen that the evaluation based on the variance of pixel intensities in the intervals is stronger against noise than the method based on the intensity difference of the intervals, which has been considered as the general method for the evaluation of similarity. The interval matching, using the cost function based on the variance, is applied to the stereo aerial photograph of an urban area, verifying its usefulness [ABSTRACT FROM AUTHOR]

This paper presents basic concepts and fundamental algorithms for processing three- dimensional digitized images (3D images). A 3D image is a pictorial datum obtained by measuring physical quantities distributed at every point in 3D space and arranging the measurements on a 3D array. First, we present several basic concepts concerning geometrical properties of 3D images, which include connectivity, the connected component, the Euler number, and deletability. Then we explain outlines of basic algorithms derived by using the forementioned concepts based on research results by the authors' group. Procedures referred to here are as follows: calculation of the Euler number, deletability test of 1-voxels, shrinking, thinning, and border following. Finally, we add 51 refer- ences which may aid more detailed research; these references also will supplement the brief explanations and small number of procedures we are limited to in this article. [ABSTRACT FROM AUTHOR]

This paper discusses the traffic flow measurement system (TMS) based on the idea of vehicle cluster. TMS is an experimental system which aims at the accurate real-time measurement of the traffic flow parameters (location of vehicle, speed, congestion, location of the queue tail, etc.) which is spread in space, using the image processing technique. As image processing techniques, the line- differentiation and the clustering based on the idea of vehicle cluster are used. By this scheme, traffic flow phenomena In the region of measurement can be measured directly. By introducing the idea of vehicle cluster, the measurement can adaptively be performed, observing the individual vehicle in low traffic and a set of vehicles in heavy traffic. As a result of the traffic flow measurement experiment using the proposed system, the following observations are made: (1) the measurement can be performed through- out the day; (2) the real-time measurement with 200-ms intervals can be performed; (3) the recognition rate for vehicle cluster is 97 percent or more for the range of measurement 150 m. The accuracy of vehicle speed measurement is ±13 percent (for the range of measurement 40 m); (5) the accuracy of vehicle height measurement is within ±10 percent (for the range of measurement 100 m an when the observation of individual vehicles is possible); and (6) the tail of the queue can be identified. [ABSTRACT FROM AUTHOR]

This paper describes a technique for guessing a complete line drawing of a polyhedron from an incomplete line drawing by an heuristic tree search. The heuristic tree search employs two indices which represent structural unity of shape: the ratio of the number of parallel sides to the number el all sides on a line drawing of a polyhedron; and the difference between the minimum `and the maximum inner angle on a loop (a closed link of sides). A line drawing obtained from preprocessing of a picture usually contains several polyhedra. It is segmented to each line drawing which corresponds to a single polyhedron, and it is generally an incomplete line drawing. The line drawing is expressed by a set of loops, and labelling on each side of the loops refers to the position and gradient of the side. An incomplete junction is searched on the line drawing by referring re the labels. Then a complete line drawing is guessed by adding sides to the incomplete junctions so that the values of the indices increase. One might expect that the present method gives aid to the recognition of an object given by an incomplete line drawing. An experimental result is given and some problems are discussed. [ABSTRACT FROM AUTHOR]

There is an increasing demand for coronary angiography, corresponding to the recent increase of such ischemic cardiac diseases as myocardial infarction. At present, the obtained angiograms must be road by specialized doctors and require a large amount of time. Still it is usually difficult to arrive at a three-dimensional understanding of the coronary artery. From such a point of view, this paper aims at the automatic analysis of the angiogram and describes a method which can determine the correspondence between vessels in the coronary artery angiograms taken along two directions. In the proposed system the three- dimensional information is used in establishing the correspondence, which is difficult for human observers to utilize. As the first step, a straight line connecting a branch point of a vessel and the X-ray source is considered and the projection of that line onto another image (called an epipolar line) is determined. Points close to that line are used as candidates for the correspondence. Then, using the vessels with established correspondences, a three-dimensional model is constructed for determination of the correct candidates. Then a prohibited region in the three-dimensional model is specified where there cannot exist a vessel. The vessels with established correspondence are represented in three dimensions and evaluation of the left ventricle is performed. From the results of experiment, the validity and the problems of the proposed system are pointed out. [ABSTRACT FROM AUTHOR]

The median filter is used frequently in image processing for smoothing while retaining edges. This paper describes an efficient method to calculate the rank filter, which is a generalization of the median filter and includes it as a special case. The presented method is simpler than the method proposed previously by Danielsson. A strict proof is given for its validity. An application to the construction of a logical circuit realizing the algorithm is also described. [ABSTRACT FROM AUTHOR]

This paper presents an outline of a prototype intelligent link system proposed to combine ‘data transmission’ and ‘data processing’ and its software system. The generative program and its processing scheme are reported in detail. The intelligent link is n pairs of a general-purpose pipe- line processing system. It is an n-stage one- dimensional line of segments which consists of a transmission line and a processor. Each segment processes received data by the programs which tag along with its data. In order to decrease the system processing time and the processor idle time, fewer bits are employed to represent a program for the information transmission. Here we use a transition matrix method to describe transition of the operations in a program. Firsts, we introduce the matrix properties and discuss applications of the matrix representation for generative programs. Secondly, we study how fewer bits can be used to represent a matrix and, as a result, find that less than 70% of the bits required to represent a matrix didrectly suffices in the proposed method. [ABSTRACT FROM AUTHOR]