In this paper, we propose a new multiple-shots transmitting method for the holographic imaging system which uses Walsh functions.
High speed imaging system using transmitting pulses modulated with a system of Walsh functions was proposed by the authors. While this system has a merit of its simple hardware, reconstructed images have many artifacts. This is regarded as a result of nonorthonormal correlations of the received wavefronts corresponding to a distinct target location.
To reduce the artifacts, we proposed a set of modulating functions which have zero cross-correlations using Hadamard matrix's characteristics.
In this method, sinusoidal signals are modulated with Walsh functions, and transmitted from multiple transmitters simultaneously. Wavefront of the echoes is detected and recorded with the array of receivers. The transmitting and receiving cycle is repeated several times. From the set of recorded waveforms, an image is reconstructed numerically.
In each cycle, the transmitters have one to one correspondence to the row numbers of an Hadamard matrix. The row number is determined with modulo 2 addition for every corresponding bits of two binary-represented numbers, the number of transmitting cycle and the number of the transmitter. We shows theoretically the property of zero cross-correlations for the set of transmitting signals.
The airborne sonar is constructed with the circular array consists of 16 transmitters and 8 receivers. The reconstructed 3-D images demonstrate the feasibility of the method.

抄録全体を表示

Automatic segmentation methods of MR images have been developed for the cardiac surgery and the brain surgery. In these fields, Region Growing method has been used mainly. In this method, the core was inserted manually, and the pixel adjoining the core was judged whether it was homogeneous or not from its features based on image information. The core grew adding the homogeneous pixels, and the region of interest was obtained as the grown core.
It is available for orthopedic surgery and biomechanics to obtain the location and the orientation of bones and soft tissues in vivo. However, MR images including them could not be segmented by the former region growing method based on only image information. This is because those tissues had fuzzy boundaries on the image.
Thus, we used not only intensity and spatial gradient as image information but also location, size and complexity of the tissue to segment the MR images. The pixel adjoining the core was judged from three local features of the pixel; its intensity, gradient and location, and two global features of the core region; its size and complexity. Judgment was performed by Fuzzy Reasoning to allow their fuzzy boundaries. The homogeneous pixel was added into the core region. It grew into normal size and smooth shape under constraint of global anatomical features.
Using the present method, as an example, radius, ulna and interosseous membrane were segmented from the multi-sliced MR images of forearm. Segmented tissues agreed with the shape inserted manually by a medical doctor. As s result, three tissues containing different features on the MR image could be segmented by a single algorithm. It takes about 10sec per slice by using an engineering workstation.

抄録全体を表示

For binocular visual space, it is well-known that horizontal lines which appear straight on a subjective frontoparallel plane are not always straight in the physical sense, and that the form of the lines depends on the distance from the observer. This phenomenon is known as Helmholtz's horopter. A similar phenomenon is also known for tactile space. Since sensory modalities are often analogous, it has been speculated that a similar phenomenon exists for auditory space, although evidence has not yet been provided of its existence. This may be due to the small number of studies on sound localization with distance which are necessary to measure the horopter for auditory space. We performed fundamental experiments on sound localization with distance and confirmed the existence of horopter phenomenon for perceived binaural auditory space. Namely:
(1) Horizontal lines which appear straight on a subjective frontoparallel plane in auditory space are not always physically straight, and their form depends on the distance between the subject and the sound sources.
(2) The dependency of the form is more similar to that for visual horopter than for tactile horopter.
(3) Although the structure of auditory space is similar to that of visual space, there exist some differences. First, the metric for auditory space and that for visual space of the same subject are not the same. Second, auditory space has more allowance than visual space for the synthesis of the information of the subjective position.

抄録全体を表示

A sensor fusion system described here extracts a target signal automatically from microphone signals corrupted by interference ambient noise. For this system, new sensor fusion algorithm for integrating audio signals and visual signals hierarchically is proposed.
In the first stage of the algorithm, audio signals and visual signals are fused each other in order to generate cue signal. Cue signal is an estimated signal of power variation of a target sound. In the second stage of the algorithm, audio signals and cue signal are fused each other to adjust the weights of an adaptive filter which extract the target signal.
A real-time processing system including 53 DSPs was constructed in order to evaluate this new algorithm. The experimental results using the real-time system demonstrates effectiveness of the cue signal which is generated by fusing audio signals and visual signals.
The advantage of proposed cue signal are discussed using numerical simulations for the first stage of this algorithm. The results of simulations demonstrates this fusion type cue signal has resistance to audio noise and visual noise, hence, it can be used in various environment.

抄録全体を表示

A new method is proposed for measuring the volume of a container using gas compressibility laws. A container whose volume is to be measured, a reference container of known volume and a bypass tube are so arranged that their electric equivalents form a bridge circuit. A loudspeaker is mounted so that one face thereof applies alternating compression to the volume undergoing measurement, while the opposite face applies compression of opposite phase to the reference volume. These compressions cause sound pressure within the two volumes and the bypass tube. And there exists within the bypass tube the balance point where the sound pressure vanishes: the bypass tube forms the ratio arms of the acoustic bridge circuit. The position of this balance point is a function of the ratio of the two volumes. Furthermore, it is affected by none of the static pressure, the gas composition, the temperature and the compression amplitude. The volume to be measured can be obtained by simple calculation from the position of the balance point. An experimental apparatus was built; that included a container whose volume could be varied from 2 to 10 [×10^-3m³] by pouring water and an electret microphone for picking up the sound pressure along the bypass tube. The proposed method has been verified experimentally with this apparatus.

抄録全体を表示

In early surgical attempts to reconstruct a neoanal function for faecally incontinent patients, the gracilis muscle was transposed around the anal canal. A constant electrical stimulation method is used to maintain muscle contraction. However, it was necessary to regulate the contraction, because too much stimulation caused fatigue or stasis of circulation to the muscle. While, too little stimulation caused incontinence. A sensor, which must function over a long time period, is needed to regulate the muscle contraction on neoanal pressure. Then we create a strain-gage pressure sensor, which is implantable between the anal canal and the transposed gracilis muscle, that measures neoanal pressure continuously.
We confirmed the following on a rabbit model;
1) This pressure sensor can be implanted between the anal canal and the transposed gradilis muscle.
2) This pressure sensor can measure neoanal pressure generated by muscle contraction.
This implantable pressure sensor can be used for the regulation of the neoanal pressure necessary for faecal continence.

抄録全体を表示

In the field of measurement and control, there are many applications where the measurement of a length of a pipe plays an important role. The present paper develops an automatic on-line measurement system which measures the length of a pipe on-line, to aim to measure the location of the time-variant sand surface in a steel pipe. This application is frequently found in the construction which improves weak foundation, irrespective sea and land. In the construction, the sand surface is fallen down in a steel pipe by repeating press and pressure reduction, which yields much water vapour in the pipe and makes the application of range sensors to the detection of the sand surface impossible.
From the viewpoint, the paper proposes to use an acoustic sensor to extract the information on the length between the sand surface and the upper end of the pipe. Furthermore, by making use of the fact that in a pipe, opened or closed at the ends, stationary waves become dominant in the fluctuation of the acoustic pressure as time passes, the paper models the fluctuation of the acoustic pressure adequately by a linear dynamic system and uses a bank of Kalman filters to estimate the length of the pipe on-line.

抄録全体を表示

In the paper (*), the realization problem of almost-linear systems which are special classes of pseudo-linear systems was presented with the following result.⌈For any invariant and affine input-response map A (equivalently, any input/output map with causality, time-invariance and affinety), there exist at least two canonical (quasi-reachable & observable) almost-linear systems which realize A And any two canonical almost-linear system with the same behavior (equivalently, input/output relation) are isomorphic each other.⌋
Moreover it is shown that the time-invariant and affine input-response map is completely characterized by a modified generalized impulse response map. And it is also shown that there exist“so-called”linear systems as examples of almost-linear system. Where“so-called”linear systems mean linear systems with non-zero initial state. Being based on the ubove results, we will investigate finite-dimensional almost-linear systems. Then we will obtain the following results.
1): a necessary and sufficient condition for finite-dimensional almost linear system to be canonical. 2): There exist uniquely a quasi-reachable standard system and an observable standard system as representatives of finite-dimensional canonical almost-linear systems. 3): a necessary and sufficient condition for the invariant and affine input-response map to be the behavior of finite dimensional almost-linear systems. 4): For a given invariant and affine input response map, there is given a realization procedure to obtain a canonical finite-dimensional almost system with the same behavior as it.
(*): Y. Hasegawa and T. Matsuo; Realization theory of Discrete-time Almost linear systems, Transaction, on SICE, 30-2 (1994)

抄録全体を表示

A parametrization of all of robust servo systems with two-degree-of-freedom is proposed by using a factorization approach, and its state space realization is derived. It is shown that the resulting systems comprise an internal model to the reference signal, an observer for the plant, a state feedback gain which stabilizes an augmented system combined with the plant and the internal model, and stable free parameters both in the feedforward path and in the feedback path. Each stable free parameter is available to improve the transient response for the reference signal input and the closed-loop properties such as sensitivity and complementary sensitivity independently. This result gives a general servo system structure which includes a well-known structure considered by Davison.

抄録全体を表示

The quality of a wide film manufactured is determined by the flatness of the film thickness along the width of the film. To control the film thickness in the lip-heater method, an uniform distribution of melted polymer flow velocity in the gap between die lips is required.
In this paper, a mathematical model of polymer flow in the lip gap is derived. The analytical model is as follows;
The flow model in the lip gap is divided into 3 sections in the flow direction because the lip gap is gradually narrowed along the flow. The flow in each section is approximated by the flow between parallel planes. The lip model is also divided into 3 sections along the flow direction and divided into the number of heaters in the direction of width in order to get a lumped model for temperature dynamics of the lip in which heaters are buried. The polymer viscosity is given by the power law equation. The polymer temperature distribution in the lip gap is obtained by the unsteady energy balance equation and the steady velocity solution. The comparison between the calculated values and the experimental data expresses that the proposed model has enough accuracy and can be used for the actual control system design.

抄録全体を表示

An interactive identification scheme is presented for remotely located objects. In the identification process, a multi-modal description is generated based on operator's instruction of the type and the image of object. The a priori information, a set of symbolic descriptions of object knowledge, is installed in the computer system on which the scheme is implemented. Supported by this a priori knowledge, the operator specifies the object in the observed image captured by monocular TV camera. The instruction is matched with observed imagery to complete the as-is object description. Through a series of experiments, the scheme has been demonstrated to be applicable to the identification of extended classes of practical scene.

抄録全体を表示

In this paper, an architecture for pattern recognition system using neural networks is presented which is insensitive to translation, scale and line thickness. The system consists of a preprocessing network and a trainable multilayered network. An input pattern is first normalized in position and size through a pattern standardizing network. A feature extraction network is next used to detect line features which are an activity pattern in the OSC (Orientation Specificity Cell) layer. The line features are also used as an input pattern to the multilayered network which is then trained to recognize the pattern. Computer simulations of a numeral recognition task show the effectiveness of the system.

抄録全体を表示

Understanding of a visual pattern information is very important in the human information processing.
In this paper, we proposed a pattern processing that relate visual patterns and symbolic knowledge. The system recognized an image by a co-working of the pattern processing and a symbolic processing. The system uses a neural network model of selective attention for the visual pattern recognition and Prolog language system for a symbol information operation.
The selective attention model was used also for a bidirectional transformation mechanism between the pattern and the symbol. This transformation mechanism enabled a reflection of the symbolic inference result to the pattern operation.

抄録全体を表示

In this paper, a method for reconstructing a 3-D shape of an object from a 2-D shading image using a Genetic Algorithm (GA), which is an optimizing technique modeled on the process of living thing's evolution, is proposed. In the proposed method, the problem of reconstructing shapes from shading images can be regarded as a kind of optimization problem, that is to search for the shape providing the most similar shading image to the shading image of the original object, and then, the GA is applied to solving the optimization problem. Each candidate of the shape corresponds to an individual in population which is formed by a lot of individuals. The fitness value of each individual is decided from the similarity of the shading image provided by the candidate of the shape to the measured shading image. The individuals with higher fitness values are selected according to mechanisms of natural selection of the GA, and then, an individual having the highest fitness value can be found finally. The 3-D shape corresponding to the individual is regarded as the reconstructed shape of the object. Using the proposed method, 3-D shapes are reconstructed from shading images of synthesized objects which are calculated by computer simulation. The results of the reconstruction demonstrate the effectiveness of the proposed method.

抄録全体を表示

A fault diagnosis system using a signed directed graph (SDG) as a qualitative model of an objective plant has been developed for the purpose of assisting operators to decide a corrective action for abnormal situations arising in a large-scale chemical plant. A lot of manpower is required, however, to construct an SDG for a chemical plant because it comprises a number of pipeline networks of complicated structure.
This paper presents a systematic method to select state variables representing a pipeline network and algorithm to synthesize its SDG utilizing the mass balance, momentum balance, heat balance and component balances. This algorithm is implemented in a computer program which synthesizes an SDG of a given pipeline network from the CAD data and is successfully applied to a test plant in order to demonstrate its usefulness.

抄録全体を表示

A learning architecture with a distributable genetic algorithm (dGA) that enables a robot system to obtain a cooperative group behavior is presented. This report focuses on the cooperation issues in a fleeing target capturing problem; mobile robots are commanded to capture fleeing targets in a 2D environment and the task requires a cooperation among mobile robots.
It is assumed that the robot system is composed of autonomous mobile robots of the same type and that each robot behaves according to its own strategies. The strategies to capture a target are encoded on the genes of each robot and dGA applies genetic operations (selection and crossover) to those genes. The difference between dGA and simple GA is that dGA carries out selection in a distributed manner. Communication is allowed only between two robots that are within each other's neighborhood and a partner for each robot is chosen through this communication.
The importance of the similarity of the gene space to the parameter space is discussed and interesting properties of both encoding and decoding method employed in this framework are shown. Issues about the degree of distribution of dGA selection are also mentioned. Simulation results show that dGA increases the number of the targets captured by the robots and, therefore, a cooperative behavior is learned by the robot system. Obtained cooperation strategies are also found to be reasonable ones.

抄録全体を表示

A new type sensor for measuring an extremely feeble grasping power by using rubber optical fiber has been developed which can measure human grasping power. In addition, we examined the possiblities to measure the patient's grasping power by a hand dynamometer.
The optical intensity of rubber optical fiber was linearly related to the applied force. The sensitivity was 5.38mV/kg and 0.56mV/kg in the range from 0 to 0.50kg and above 0.50kg, respectively.

抄録全体を表示

It is well known that the minimality of the discretized system is not preserved in the specific sampling period. We have investigated the quantitative properties of the discretized system with he neighborhood of the specific sampling period.

抄録全体を表示

In this paper, we propose a synthesis approach of nonlinear state observer for inverted pendulum systems based on the quadratic stabilization problem and confirm its validity by simulations.

抄録全体を表示