In this paper, we introduce a plan of interrelationship mining using rough sets. In general, decision rules as results of rough set-based data analysis describe characteristics among objects that are representable by combinations of attributes and those values. Interrelationship mining we propose enables us to extract characteristics among objects with respect to interrelationship between condition attributes by comparing attribute values of different condition attributes.

View full abstract

This article reports an application of Rough Non-deterministic Information Analysis (RNIA) to two data sets. The one is Mushroom data set in UCI machine leaning repository, and the other is students' questionnaire data set. Even though these data sets include lots of missing values, we obtained some interesting rules by employing NIS-Apriori algorithm to these data sets.

View full abstract

In this paper, we propose a revised method of rule extraction from decision tables using logic minimization technique. Rule extraction using logic minimization technique has been proposed as one of the method of heuristic rule extraction from decision tables. The authors have improved the method of rule extraction from large scale decision table, by improving the algorithm of expansion process. As a result, we can reduce redundant rule but calculation time increased. In this paper, we try to reduce calculation time by improving the algorithm further. We also verify the effectiveness of proposed algorithm comparing it with existing algorithm.

View full abstract

In this paper, we discuss how to select parameters of a multi-agent space in the artificial Kansei system based on a variable neighborhood model, using the language image scale proposed by NIPPON COLOR & DESIGN RESEARCH INSTITUTE INC.

View full abstract

In this research work, the author tries to generate several pairs of a t-norm and a t-conorm on a partially ordered set/a lattice in the framework of BCK-algebras. As the concrete model of such a system, a theory of multisets equipped with set operations based on discrete t-conorms and discrete t-norms is presented.

View full abstract

In rough set approaches, rules infering the membership of a single class have been induced, while rules infering the membership of a union of classes can also be induced in the same way. It is demonstrated that the classifier based on rules with unions of classes outperforms that based on single class rules. In this paper, from this fact, we investigate the way of building the classifier with a better performance based on rough set rule induction. Through numerical experiments, we show the effectiveness of union rule induction and the multiplication of explicable rule sets.

View full abstract

In this study, we propose a new decision rule induction approach. Conventional rule induction methods are based on sequential covering with the general-to-specific approach in which to generate a premise of a rule, the premise is initialized to be empty and conditions are added to it until no or few negative objects are covered by the premise. While, in this study, we propose a rule induction method using the specific- to-general approach by applying discernibility based clustering to rule induction. In our approach, positive objects are clustered using a cohesion measure which is related to discernibility of clusters. From an obtained cluster, we can generate a premise of a decision rule by the common condition values of objects in the cluster.

View full abstract

We consider some properties of granular hierarchical structure of monoids.

View full abstract

This article presents a formulation of rough multisets by introducing upper and lower approximation into multisets in terms of commutative diagrams in granular hierarchical structure.

View full abstract

A theoretical framework of rule induction based on rough sets is shown in possibilistic information system. Each object with non zero possible membership degrees possibly supports rules. On the other hand, every object with non zero certain membership degrees do not certainly support rules. We cannot know what rules are induced from rough approximations without considering characteristic value of objects. This leads to our formulating rough approximations under considering characteristic values of objects. Furthermore, we have introduced a criterion for valuable rules. Our approach is free from the restriction that possibilistic information of attributes in possibiilistic information systems.

View full abstract

As the technique to analyze the causal relationship of categorical data with few samples, rough sets is beginning to spread in the field of product planning or a product design. There is high request of desiring to use this method in the early stage of development. For that purpose, it can be necessary to extract the decision class of rough sets by few subjects. Then, we have proposed method of presuming the decision class by few subjects. As the result of verifying using examples, the validity of the proposal method was shown in this paper.

View full abstract

Fuzzy c -Means (FCM) has been applied to very large data sets composed of a huge number of objects based on several sequential sampling approaches. In this research, the sequential sampling approaches are applied to an FCM-type co-clustering model of FCCM, and its applicability to very large cooccurrence relational data composed of a huge number of objects such as user-item purchase history data is demonstrated by revealing the intrinsic co-cluster structures among objects and items.

View full abstract

Cluster Validation is an important mechanism in FCM-type clustering applications with the goal being to select optimal cluster numbers. Xie-Beni index is an established validity measure, in which cluster optimality is measures considering cluster compactness and separateness. In this research, a Xie-Beni-like validity index is applied to validation of fuzzy co-clusters, and its availability is considered in very large data applications.

View full abstract

Collaborative filtering in Internet societies is a promising application area of co-clustering. In order to search for the items to be recommended to an active user, co-cluster structures of users and items are considered. With the goal of applying real world applications with very large users, the conventional clustering algorithm suffers from huge computational costs and must be modified for reducing the costs. In this research, the applicability to very large data is studied in conjunction with some sampling approaches.

View full abstract

Most clustering methods focus on numerical data. However, lots of the data existed in the databases are both categorical and numerical. Until now, clustering methods have been developed to analyze complete data. Although we sometimes encounter data sets which contain missing one or more feature values (incomplete data), traditional clustering methods cannot be used for these kinds of data. Therefore, we study this problem and discuss clustering methods which can handle mixed numerical and categorical incomplete data. Further, we apply fuzzy clustering for interpreting the result with vagueness.

View full abstract

In this report, a maximization-based Bezdek-like Fuzzy c-Means clustering methodology is proposed. Standard fuzzy c-means and its many variants are based on minimization model, which contrast with a method based on maximization model derived from entropy-regularization of spherical K-means. This report shows the fuzzification parameter less than 1 yields maximization-based Bezdek-like Fuzzy c-Means clustering method. Following to this manner, a maximization-based fuzzy nonmetric model and a maximization-based Bezdek-like fuzzy co-clustering model are proposed. Furthermore, some kernelized maximization-based Bezdek-like fuzzy c-means clustering algorithms are proposed, two of which have eigenvalue subproblem.

View full abstract

Recently, K-member clustering method attracts many researcher’s interest in the field of the privacy protection. The method automatically classifys many objects into some clusters of which the size is more than K, however, this problem is known as NP-complete. This paper proposes Extended Two-Division Clustering for K-Anonymity of Cluster Maximization and verifies the clustering methods through nermerical samples.

View full abstract

Clustering is one of the method of data analysis. Rough k-means (RKM) by Lingras et al. is one of rough clustering algorithms[3]. The method doesn’t have a clear indicator to determine the most appropriate result because it is not based on any objective functions. Therefore, a rough clustering algorithm based on optimization of an objective function was proposed[6]. This paper will propose a new rough clustering algorithm based on optimization of an objective function with fuzzy-set representation to obtain better lower approximation and estimate the effectiveness through some numerical examples.

View full abstract

Possibilistic clustering is well-known as one of the useful techniques because it is robust against noise in data. We propose entropy based possibilistic clustering with L1-regularization and show its classification function. We, moreover, show the result of sequential extraction by proposed method.

View full abstract

A family of fuzzy neighborhood models are introduced which takes mutual distances of keywords in a text, whereby keyword space becomes a kernel Euclidean space. Therefore many methods based on kernel spaces can be applied. Theory and applications to web text are discussed.

View full abstract

Spectral clustering is based on the graph Laplacian, and outputs good results for well-separated groups of points even when they have nonlinear boundaries. However, it is generally difficult to classify a large amount of data by this technique because computational complexity is large. We propose an algorithm using the concept of core points in DBSCAN. This algorithm first applies DBSCAN for core points and performs spectral clustering for each cluster obtained from the first step. Efficiency of the proposed algorithm is shown by the analysis of complexity. Simulation examples are used to show performance of the proposed algorithm.

View full abstract

Spectral clustering is based on the graph Laplacian, and outputs good results for well-separated groups of points even when they have nonlinear boundaries. However, it is generally difficult to classify a large amount of data by this technique because computational complexity is large. We propose an algorithm using the concept of core points in DBSCAN. This algorithm first applies DBSCAN for core points and performs spectral clustering for each cluster obtained from the first step. Efficiency of the proposed algorithm is shown by the analysis of complexity. Simulation examples are used to show performance of the proposed algorithm.

View full abstract

Scheduling of semiconductor wafer testing processes can be seen as a resource constraint project scheduling problem (RCPSP). However, it includes uncertainties caused by human factors, wafer errors and so on. Because some uncertainties are not simply quantitative, the range estimation of the parameters would not be very useful. Considering such uncertainties, finding the meta rule to select a dispatching rule depending on the situation would be more suitable than solving the RCPSP under uncertainties. Moreover, this meta rule would be advantageous in the adaptation to unexpected changes. In this paper we apply some machine learning approaches to acquire the meta rule selecting a dispatching rule depending on the situation. We compare the obtained rules with the simple dispatching rules and examine the effectiveness and usefulness of the obtained rules in the problems with unpredictable wafer errors.

View full abstract

In mathematical programming problems, it is often to apply fuzzy numbers in order to represent uncertainty of parameters. The appropriateness of membership functions is directly related to that of decision making. In this talk, we discuss an appropriate constructing approach of membership functions in decision making.

View full abstract

This article considers bilevel linear programming problems where the coefficients of the objective functions and the constraints in the problem are given as fuzzy parameters. Stackelberg solutions under fuzziness are dened by incorporating the notions of possibility theory into the original concept of Stackelberg solutions. It is shown that Stackelberg problems under fuzziness are transformed into deterministic bilevel linear or nonlinear programming problems, and that the resulting problems are exactly solved by using conventional bilevel linear or nonlinear programming techniques.

View full abstract

Focusing that the equation of the exponential smoothing method(ESM) is equivalent to (1,1) order ARMA model equation, a new method of the estimation of the smoothing constant in exponential smoothing method was proposed before by us which satisfies the minimum variance of forecasting error. In this paper, we utilize the above stated theoretical solution. Firstly, we estimated the ARMA model parameter and then estimate the smoothing constants. Thus the theoretical solution is derived in a simple way and it may be utilized in various fields. The new method shows that it is useful for the time series that has various trend characteristics.

View full abstract

In industry, making a correct forecasting is a very important matter. If the correct forecasting is not executed, there arise a lot of stocks and/or it also causes lack of goods. Time series analysis, neural networks and other methods are applied to this problem. In this paper, neural network is applied and Multilayer perceptron Algorithm is newly developed. The method is applied to the Airlines Passengers and Cargo Data. When there is a big change of the data, the neural networks cannot learn the past data properly, therefore we have devised a new method to cope with this. Repeating the data into plural section, smooth change is established and we could make a neural network learn more smoothly. Thus, we have obtained good results. The result is compared with the method we have developed before. We have obtained the good results.

View full abstract

In recent years, severe competition is executed both on getting air passengers and those of air cargos. The forecast of the number of taking-off and landing is expanding, while demand for air cargo is decreasing. Strict marketing is required in such fields. Forecasting the trend of air cargo is an essential item to be investigated in airlines. In order to make forecast for time series, the method of using linear model is often used. Forecasting using neural network is also developed. Reviewing past researches, there are many researches made on this. There is many room to improve in neural network, therefore we make focus on them. We use time series data, and in order to make forecast, a new coming data should be handled and the parameter should be estimated based upon its data. This is a so-called on-line parameter estimation. In this paper, neural network is applied and Multilayer perceptron Algorithm is newly developed. The method is applied to the Airlines Passengers and Cargo Data in the case of Weekly data. When there is a big change of the data, the neural networks cannot learn the past data properly, therefore we have devised a new method to cope with this. Repeating the data into plural section, smooth change is established and we could make a neural network learn more smoothly. Thus, we have obtained good results. The result is compared with the method of ARIMA model. We have obtained the good results. The new method shows that it is useful for the time series that has various trend characteristics and has rather strong seasonal trend. The effectiveness of this method should be examined in various cases.

View full abstract

In this paper, we performed several simulations based on a condition that a robot and a human have long-term interactions in our proposed model. This model generates several emotions for robots that consider mutual effects of desires and emotions because our earlier model remains imperfect in that the robots express the same emotions when they receive an external-input from a user. In the model, we express the mutual effects of desires and emotions on the basis of the internal-states of a robot, such as physiological factors. These internal-states of a robot are contained by robots, which possess their current states. In addition, the model generates emotions and desires by using two self-organizing maps (SOMs). The simulation results show that the proposed model, which gets biased inputs in growth process, gets biased differentiation states and expresses emotions which are influenced by the growth process．

View full abstract

In this paper, we present a care action generation model for communication between robots with emotion growth function. Hence, we propose emotion generation model for robot with emotion growth function for smooth communication between human and robot. Robots have to communicate with emotion like creature, in order for human to feel familiar to the robot. However, robots of conventional model were unsolved functions of communication like creature. Hence, to solve this problem, we propose a care action generation model that enables communication like creature a parent robot and a child robot with growth functions interact with each other. The parent robot has grown by mechanical input. In contrast, the child robot has grown by the parent robot, without the human input. Through mutual interaction of the parent robot and the child robot, the child robot gets grown and the parent robot grows. We run simulations with proposed model to verify which is effective for generation of care action like creature. In simulation results, we confirm that care actions of the parent robot are effective in getting rid of desire of the child robot.

View full abstract

Recently, various robots have been used in the human life, and it is thought that the interaction of human and robots increases. In that case, it is important to model the Kansei that human feels when the action of robot is seen. However, it is difficult to define the action of robot such as “pretty”. In this paper, we try to analyze relation between robot actions besed on “Large-Small” and “Fast-Slow” and Kansei words. It is easy for humans to image these actions.

View full abstract

Recently, the robot technology which can realize the smooth communication with human has been required. In this laboratory, we aim to realize Interactive Emotion Communication (IEC) - which is a bidirectional communication between human and robot. But we have a problem that we obtained the subject’s body features only in the moment in our previous studies, therefore, the robot could not recognize the continuous motion of subject sufficiently. In order to solve this problem, in this research, we propose a method which the robot enables to learn the continuous pattern of human’s body motion in beforehand, by using the recurrent neural network, which can learn the time series information. Based on this method, we build a system that robot can recognize the continuous motions intended by human in real time.

View full abstract

Recently, the applications of robots, especially in social service and daily communications, have been widely spread. In order to make these robots and humans live close together, it is important for robots to increase the personal affinity and to have human-like qualities. For example, if we can give a similar to human character to robots and the robots can adapt themselves to the character of humans, we think that the personal affinity will increase greatly. Therefore, in this research, we build a robot to have complex and various characters, and we aim at the establishment of method which is adaptable for human characters. In this case, robot learns human characteristics from interactions between robot and humans, we construct a system that the robot expresses the personality which fits for its characteristics by themselves.

View full abstract

Abstract In this paper, we verify an effectiveness of “Kansei rules” to analyze impressions of products. In our previous paper, we have proposed a method for making Kansei rules by using a parallel Genetic Algorithm (GA). The Kansei rules are generated by results of questionnaires. These results determine users’product impressions and help to uncover major factors that affect product’s attractiveness. Therefore, we analyzed appearance of running shoes and investigated the factors that determine particular user preferences. In this paper, we examine to estimate consumer’s overall impression of the product from their response to particular components of unknown questionnaire data by using generated rules. In the experiment, we compared performance of rules by using a parallel GA and by using a fuzzy C4.5 decision tree of our previous study. As a result, we verified the effectiveness of the method by using a parallel GA.

View full abstract

Decision theory with a payoff matrix is useful for selecting an alternate under uncertain state. The payoff matrix denotes estimate profit or loss obtained by each alternate under each state of nature. Some selection criteria have been proposed already in order to select an alternate. However, we can use several unused mathematical operations as the selection criteria. In this article, we will propose some novel selection criteria and also describe affects of negative payoff values.

View full abstract

Using relative individual scores among alternatives, in this paper, we propose a set function that shows the feature of an alternative. In a weighted sum model such as the Analytic Hierarchy Process (AHP), a set function value is constructed from the weights of the model. The set function value for the alternative is calculated by averaging the values of the set function representation of randomly generated weights when the alternative has the highest comprehensive score. By interpreting the functions, we can understand the feature of an alternative. We show the properties of the set functions and the sensibility analysis.

View full abstract

Research participants of this research were 50 young people and 50 elderly people. The participants were asked to evaluate the degree of proximity between themselves at present and themselves at age 0 to 100 with the multi-axis concentric circle scale. The participants were also asked to estimate the degree of satisfaction with life, happiness, confidence in health, optimistic tendency and volition that they had about themselves at age 0 to 100. The results showed that the differences in change patterns of the mean scores of these evaluation items at 12 age categories between the two different-aged groups were statistically significant. More effective application methods and problems with time recognition in the life-span developmental process using the multi-axis concentric circle scale were discussed.

View full abstract

This paper, noting skylines comprised of roof shapes, proposes desirable cityscapes for two types of groups one of which consists of subjects who attach importance to uniformity and the other of which consists of subjects who does not. A comparison is made between a stimulus (Case 2) in which roof shapes vary over town blocks and a stimulus (Case 3) in which roof shapes vary over and within town blocks. The inferred evaluation score of each time period implies that synergistic effect of sceneries in town blocks arises in both stimuli. Furthermore, the average of evaluation scores across subjects who attach importance to uniformity show the superiority of the stimulus of Case 2.

View full abstract

Brain-Computer Interface (BCI) is a system that controls external devices based on the signal of human’s brain. The P300 speller, which uses P300 as the target feature, is one of the BCI communication tools. P300 speller allows a user to select letters just by user’s thought. However, generally a user can not switch P300 speller ON and OFF by himself/herself. This paper proposes and studies a P300 speller which is synchronous to the intention of input of user based on the probability density distribution of discriminant score.

View full abstract

The authors measured electroencephalograms (EEGs) from subjects looking at words hiragana (Japanese alphabet), and recalling one character Kanji homophone. The words are presented to the subject at random. Each word consists of some Hiragana characters, and has some one character Kanji homophones. The equivalent current dipole source localization (ECDL) method is applied to the event related potentials (ERPs): summed EEGs. ECDs are localized to the primary visual area V1, to the ventral pathway (ITG: inferior temporal gyrus), and to the Broca's area, etc. In addition ECDs are localized to the hippocampus, to the parahippocampal gyrus (ParaHip) and to the fusiform gyrus (FuG), etc. We compared the difference in localized results between subjects.

View full abstract

The authors measured electroencephalograms (EEGs) from subjects in imaging or recognizing several types of images presented on CRT. Each image consists of four types of line drawings of body part, tetrapod, home appliance, etc. The canonical discriminant analysis is applied to the single trial EEGs. Four channels of EEGs at the right frontal and temporal are used in the discrimination. They are Fp2, F4, C4 and F8 according to the international 10-20 system. Sampling EEGs were taken from 400ms to 900ms at 25ms intervals. Also, data were resampled -1ms and -2ms backward. The number of variates is twenty one by four; so the data were eighty four dimensional vectors and number of the data was three hundred and sixth. Results of the canonical discriminant analysis by use of so called jack knife method were greater than 95 %. These results are improved from the precedent our research.

View full abstract

Visions of smart houses and home automation technologies have been around for over three decades. Since that time, computers and technology have made a huge step forward and simple home automation is not so appealing anymore. In this paper we propose and prototype an intelligent living room system that without intended interaction enhances people everyday life by figuring out our desires from our natural gestures, facial expressions and speech. Human behavior such as gestures and facial expressions but also preferences vary depending on the person and environment. The meaning of the same gesture and facial expression made by different persons, or by the same person but in different situation, can have a different meaning. Therefore the intelligent system must be able to recognize the person and the situation and learn the preferences. In order to achieve this, artificial intelligence and machine learning algorithms, such as Hidden Markov Model (HMM) and Growing Self-Organizing Map (GSOM) are used. Microsoft Kinect for Windows sensor is used to monitor gestures, voice and locate people in the living room. High definition camera is used in detecting facial expressions. The information gathered from multiple sensors and users' desires recognized from gestures and facial expressions are combined in order to make correct decisions. As a result, the system seamlessly enhances people everyday life by making it more comfortable by, for example, changing the temperature, room lighting setting and ventilation. The system will learn each individual's preferences in different situations. It will adapt to different users and take actions based on user's postures, gestures, speech and facial expressions and also location in the room.

View full abstract

We discuss how to indicate the logicality and connectivity from living neuronal network in vitro. Rat hippocampal neurons are organized into complex networks in a culture dish with 64 planar microelectrodes. We have already proposed a model to analyse logic of signals of three electrodes in a culture dish with t-norm and t-conorm connectives. However, in this paper, we expand the electrode space which is analyzed, and discuss the logicality and connectivity of cultured neuronal network from the view of signal transfer with propagation, spread and convergence. First, we define the transfer pattern of pulse signal with formulation of propagation, spread and convergence. Next, the connectivity of electrodes is defined by contribution of fuzzy sets which represents signal strength. The useful of the model is evaluated by experiment data of cultured neuronal network.

View full abstract

For the interface between a human neuronal network and a machine circuit, it is critical to establish the neuronal decoding technology. The information processing of the neuronal network is embedded in network dynamics. The dissociated rat hippocampal neurons on multi-electrodes array dish are useful as a simple model of brain information processing system. In this research, we extracted pattern repertory from the spontaneous electrical activity in living neuronal network (LNN). One of the clustering methods, X-means with kkz preprocessing, was applied to the series of feature vectors of numbers of spikes within a time window at 64 electrodes. The width of time window was set to 5 ms, which include only a spike at the most. The X-means method estimates the adequate number of clusters according to Bayesian information criterion, however, the number of clusters in 30 min recorded data was estimated as extremely large, approximately over 1000. Such too sensitive clustering results are considered to be because of the uniformity of the weight of feature vector elements. However, limiting to clusters with at least 1% occurrence, approximately only 15 pattern repertories (clusters) repeatedly occurred. In addition, the cycle of the pattern repertory was approximately 30-40 s.

View full abstract

In order to realize the interface between a neural network and an electronic circuit for neuroprosthetics, it is critical to elucidate the network dynamics of a neural network and to establish the optimal input methods and the decoding methods of the spatio-temporal firing pattern of the neural network. For the purpose, we are developing the neuro-robot as the model system for biological information processing with vital components. The behavior of the neuro-robot is generated by the response pattern of neuronal electrical activity evoked by a current stimulation as an input from outer world. Our developed neurorobot has a living neuronal network (LNN) as an epistatic processor and the robot is designed to generate purposive behavior by the regulation of the relationships between input and output of the LNN. In this study, we developed a novel type of neuro-robot with Self-Organization Map (SOM) algorithm, which enable the robot to perform non-stop learning and generation of behavior simultaneously. The 64 dimension feature vectors of the spatiotemporal electrical pattern evoked by the inputs according to the value of the IR sensors on the robot body are inputted to the SOM. The 64 dimension feature vectors are mapped to the 10 x 10 output layer of SOM. Only at the early stage of the learning, SOM selects two winner nodes premisely assigned to specific inputs for the obstacles near the L and R side of the robot body. Thus the process is teacher learning. Winner nodes are linked to the purposive behaviors adequate to the inputs. We call the process as "seeding". By seeding procedure, the distribution of winner units for the two inputs were separated each other, in the case of the spatiotemporal pattern of responses evoked by 2 inputs were not overlapped. In addition, the location of the center of gravity among winner nodes for repeated inputs are approximately same in the output layer of the SOM.

View full abstract

We have developed ``Air Brain", a telemetric system for EEG and human behavior. The advantage feature of this system is that it enables us to measure the brain waves and human behavior easily, anywhere and anytime. The system is composed by a smartphone and a compact EEG made by general-purpose parts, thus the system can be easily introduced at low-cost. The Wearable EEG is connected to a smartphone via Bluetooth communication. The data has been transported to the data server on Internet via 3G-Network of smartphone, enabling us telemetry on the broadest area at present. The recent up-to-date smartphone is equipped with various high performance sensors and GPS. Utilizing these apparatus for sensing human behavior, Air brain system simultaneously offers details of human behavior and EEG at that time. We have confirmed that Air brain system detects alpha-waves form the electrode on a fore head, equally as a commercial EEG amplifier. Moreover, using the Air Brain system, we found that alpha waves become to be dominant immediately after walking.

View full abstract

In recent years, development BCI, make an interface between machine and our brain, is widely conducted using mostly EEG. Although EEG may be successful for BCI systems, EEG measures brain wave activity that humans can’t control, so the subject needs training to use the EEG for BCI a long time. Therefore, in this study, NIRS was implemented to develop a BCI system that doesn’t require long practice time. We used a SVM classifier system that presumed the onset and offset of the task of both the left and right hand from a NIRS signal during finger motion.

View full abstract

In recent years, with the recent release of hybrid camera combining RGB camera and depth sensors, the accuracy of motion analysis has greatly improved. On the other hand, research areas such as ethology, or study of animal behavior, has great demand for automatic image analysis and detection using video cameras. For this research, we developed an automatic tracking system for small animals using a hybrid camera and verified the effectiveness of the developed system by experiment using gerbils.

View full abstract

Giving a feel to a virtual object is very effective, in that it can be improved in operability and reality. Recently, giving a sense of touch by using a visual illusion has been attracting attention, as a way to be given a sense of touch without going through the equipment. However, it has not been quantified about the relationship of tactile and visual, and there is a room for study in its augmentability. In this study, we investigate how much the vision contributes to the tactile. In addition, we propose three methods for providing a new finger tactile sense only by augmentation of visual information.

View full abstract

It is convert to document papers or books into digital image data using a camera. The image shows the inclined target and wasted background regions as the image is not obtained carefully from the front. In order to strip off discarded area and to clarify information on the target area, the projection transformation is utilized to the target segment. This method, however, presuppose that the target is on the planar surface, so when the target is on the curved surface, distorted image is generated. Therefore we propose the method to estimate the target curve surface, and to expand planar surface suitable for the projection transformation.

View full abstract