Keiji Gyohten | Oita University (original) (raw)

Papers by Keiji Gyohten

Proceedings of the 9th International Conference on Pattern Recognition Applications and Methods, 2020

2020 Computing in Cardiology Conference (CinC), 2020

Most traditional studies on arrhythmia detection in electrocardiogram (ECG) have proposed general... more Most traditional studies on arrhythmia detection in electrocardiogram (ECG) have proposed general methods applicable to various patients. Since patients have their own unique ECG patterns, it becomes possible to detect abnormalities that could not be found by the general methods if we could propose a new arrhythmia detection method tailored to each patient. Furthermore, the new method can effectively support doctors in their diagnosis if it could give the basis for determining the abnormality. In this study, we propose an individualized ECG abnormality judgment method using Autoencoder and convolutional neural network (CNN). This method makes Autoencoder learn only normal waveforms that can be easily collected enough and obtains the characteristics of the individual's unique normal waveforms. Our method compares the features acquired from ECG to be analyzed with those of the normal waveform and determines whether they are normal or abnormal. In addition, we aim to construct a system that can show the basis of the judgment whether it is normal or abnormal by showing the acquired features.

Proceedings of the 9th International Conference on Pattern Recognition Applications and Methods, 2020

In this research, we propose a method to identify the cause of misrecognition in offline handwrit... more In this research, we propose a method to identify the cause of misrecognition in offline handwritten character recognition using a convolutional neural network (CNN). In our method, the CNN learns not only character images augmented by applying an image processing method, but also those generated from character models with stroke structures. Using these character models, the proposed method can generate character images which lack one stroke. By learning the augmented character images lacking a stroke, the CNN can identify the presence of each stroke in the characters to be recognized. Subsequently, by adding dense layers to the final layer and learning the character images, obtaining the CNN for the offline handwritten character recognition becomes possible. The obtained CNN has nodes that can represent the presence of the strokes and can identify which strokes are the cause of misrecognition. The effectiveness of the proposed method is confirmed from character recognition experiments targeting 440 types of Japanese characters.

SMC 2000 Conference Proceedings. 2000 IEEE International Conference on Systems, Man and Cybernetics. 'Cybernetics Evolving to Systems, Humans, Organizations, and their Complex Interactions' (Cat. No.00CH37166), 2000

The paper describes a method for understanding an image where desired objects have part-of relati... more The paper describes a method for understanding an image where desired objects have part-of relationships between them. This method is based on a hierarchical multi-agent system, where each agent takes charge of a desired object and tries to extract it using knowledge on its features. Since users can define this knowledge freely without any modification of the algorithm, this method

2020 Computing in Cardiology Conference (CinC), 2020

Most traditional studies on arrhythmia detection in electrocardiogram (ECG) have proposed general... more Most traditional studies on arrhythmia detection in electrocardiogram (ECG) have proposed general methods applicable to various patients. Since patients have their own unique ECG patterns, it becomes possible to detect abnormalities that could not be found by the general methods if we could propose a new arrhythmia detection method tailored to each patient. Furthermore, the new method can effectively support doctors in their diagnosis if it could give the basis for determining the abnormality. In this study, we propose an individualized ECG abnormality judgment method using Autoencoder and convolutional neural network (CNN). This method makes Autoencoder learn only normal waveforms that can be easily collected enough and obtains the characteristics of the individual's unique normal waveforms. Our method compares the features acquired from ECG to be analyzed with those of the normal waveform and determines whether they are normal or abnormal. In addition, we aim to construct a system that can show the basis of the judgment whether it is normal or abnormal by showing the acquired features.

2020 Computing in Cardiology Conference (CinC), 2020

1.1 Background With the recent enhancement of desktop design environments, it has become easy for... more 1.1 Background With the recent enhancement of desktop design environments, it has become easy for personal users to design graphical documents such as posters, flyers, slides, drawings, etc. These kinds of documents are usually produced by the applications like drawing softwares, which have the advantage that they can store and retrieve the drawing data electronically. By reusing parts of the stored drawing data, the users can design the graphical documents much more easily. However, generally, the stored data of many users is not shared, although this can be achieved by putting a drawing database. One reason is that it is difficult to retrieve desired figures from large amounts of drawing data in the database. Unlike in text search, the figure search will require enormous amounts of computation time because matching of the geometric primitives in the drawing data will cause their combinatorial explosion in 2D space. To address this problem, many approaches have been proposed recent...

Technical report of IEICE. PRMU, 2009

Inthis paper, wepropose asystem whichdecides camera workina3Dvirtual soccer gameandgenerates 3DCG... more Inthis paper, wepropose asystem whichdecides camera workina3Dvirtual soccer gameandgenerates 3DCGanimation automatically, called DataAnimating. The3Dvirtual soccer gameisrepresented bytheplayevents including player IDs, player positions, types ofplayandsoon.These properties arecreated byreal soccer video analysis system, called Digital Scorebook(1). Thesystem generates a3DCGanimation according totwouserselections ofsoccer playevents. Tocreate impressive soccer animation fromreal soccer game,wefocus onvirtual cinematography asthe artofrhetoric ofanimation scene. Thesystem canautomate acomposition ofanimation story anddecide camerawork based onthevirtual cinematography. Ouranimation story hasthethree part, Introduction, Development andConclusion, selected fromtheorganization ofChinese poetry. Theappropriate cameraworkisdecided ineachpart. Moreover, the picture composition foreachpart inascreen iscalculated based ongolden ratio. Intheexperimental result, thesystem cangenerate 3DCGanima...

IEEJ Transactions on Electronics, Information and Systems, 2020

IEEJ Transactions on Electronics, Information and Systems, 2016

It is well known that points on a plane in 3D world are related to corresponding image points in ... more It is well known that points on a plane in 3D world are related to corresponding image points in a view of a moving camera by projective translation. Good image features have robust projectivity under any camera movements. In the standard performance evaluation of image processing, real captured images of a scene are used ordinarily. However, it is not enough to evaluate in detail because the variation of camera angle and distance to target objects are limited and the capturing cost is expensive. During the early stage of the image processing development, the basic performance measurement should be the most important in an easy way. We propose a projectivity diagnosis method to measure the performance of local descriptor base template matching between a template image and reference images which are created by deforming the template image. This template matching consists of a feature image point extraction and a local descriptor matching. The proposed method evaluates the positional accuracy of the extracted feature points and the matching with local descriptor. Four metrics are introduced to evaluate the projectivity of template matching. In the experiment, our proposed diagnosis method expose the projectivity of SIFT, SURF, and ORB. SIFT showed the better robustness than the others.

2018 16th International Conference on Frontiers in Handwriting Recognition (ICFHR), 2018

This paper presents a method which reduces uncertainty of a position and a direction of an autono... more This paper presents a method which reduces uncertainty of a position and a direction of an autonomous robot by observing environment with a camera. In the proposed method, the state of the robot is represented by a state vector obeying a probability distribution. The robot creates and renews an environment map by considering the information from the mounted camera. Moreover, the robot revises the probability distribution on its state by comparing the map and the information from the camera. This attempt at avoiding the inconsistency in the map reduces the uncertainty of the state of the robot. We present experimental results to show the possibility of this method.

Proceedings of the 9th International Conference on Pattern Recognition Applications and Methods, 2020

2020 Computing in Cardiology Conference (CinC), 2020

Most traditional studies on arrhythmia detection in electrocardiogram (ECG) have proposed general... more Most traditional studies on arrhythmia detection in electrocardiogram (ECG) have proposed general methods applicable to various patients. Since patients have their own unique ECG patterns, it becomes possible to detect abnormalities that could not be found by the general methods if we could propose a new arrhythmia detection method tailored to each patient. Furthermore, the new method can effectively support doctors in their diagnosis if it could give the basis for determining the abnormality. In this study, we propose an individualized ECG abnormality judgment method using Autoencoder and convolutional neural network (CNN). This method makes Autoencoder learn only normal waveforms that can be easily collected enough and obtains the characteristics of the individual's unique normal waveforms. Our method compares the features acquired from ECG to be analyzed with those of the normal waveform and determines whether they are normal or abnormal. In addition, we aim to construct a system that can show the basis of the judgment whether it is normal or abnormal by showing the acquired features.

Proceedings of the 9th International Conference on Pattern Recognition Applications and Methods, 2020

In this research, we propose a method to identify the cause of misrecognition in offline handwrit... more In this research, we propose a method to identify the cause of misrecognition in offline handwritten character recognition using a convolutional neural network (CNN). In our method, the CNN learns not only character images augmented by applying an image processing method, but also those generated from character models with stroke structures. Using these character models, the proposed method can generate character images which lack one stroke. By learning the augmented character images lacking a stroke, the CNN can identify the presence of each stroke in the characters to be recognized. Subsequently, by adding dense layers to the final layer and learning the character images, obtaining the CNN for the offline handwritten character recognition becomes possible. The obtained CNN has nodes that can represent the presence of the strokes and can identify which strokes are the cause of misrecognition. The effectiveness of the proposed method is confirmed from character recognition experiments targeting 440 types of Japanese characters.

SMC 2000 Conference Proceedings. 2000 IEEE International Conference on Systems, Man and Cybernetics. 'Cybernetics Evolving to Systems, Humans, Organizations, and their Complex Interactions' (Cat. No.00CH37166), 2000

The paper describes a method for understanding an image where desired objects have part-of relati... more The paper describes a method for understanding an image where desired objects have part-of relationships between them. This method is based on a hierarchical multi-agent system, where each agent takes charge of a desired object and tries to extract it using knowledge on its features. Since users can define this knowledge freely without any modification of the algorithm, this method

2020 Computing in Cardiology Conference (CinC), 2020

Most traditional studies on arrhythmia detection in electrocardiogram (ECG) have proposed general... more Most traditional studies on arrhythmia detection in electrocardiogram (ECG) have proposed general methods applicable to various patients. Since patients have their own unique ECG patterns, it becomes possible to detect abnormalities that could not be found by the general methods if we could propose a new arrhythmia detection method tailored to each patient. Furthermore, the new method can effectively support doctors in their diagnosis if it could give the basis for determining the abnormality. In this study, we propose an individualized ECG abnormality judgment method using Autoencoder and convolutional neural network (CNN). This method makes Autoencoder learn only normal waveforms that can be easily collected enough and obtains the characteristics of the individual's unique normal waveforms. Our method compares the features acquired from ECG to be analyzed with those of the normal waveform and determines whether they are normal or abnormal. In addition, we aim to construct a system that can show the basis of the judgment whether it is normal or abnormal by showing the acquired features.

2020 Computing in Cardiology Conference (CinC), 2020

1.1 Background With the recent enhancement of desktop design environments, it has become easy for... more 1.1 Background With the recent enhancement of desktop design environments, it has become easy for personal users to design graphical documents such as posters, flyers, slides, drawings, etc. These kinds of documents are usually produced by the applications like drawing softwares, which have the advantage that they can store and retrieve the drawing data electronically. By reusing parts of the stored drawing data, the users can design the graphical documents much more easily. However, generally, the stored data of many users is not shared, although this can be achieved by putting a drawing database. One reason is that it is difficult to retrieve desired figures from large amounts of drawing data in the database. Unlike in text search, the figure search will require enormous amounts of computation time because matching of the geometric primitives in the drawing data will cause their combinatorial explosion in 2D space. To address this problem, many approaches have been proposed recent...

Technical report of IEICE. PRMU, 2009

Inthis paper, wepropose asystem whichdecides camera workina3Dvirtual soccer gameandgenerates 3DCG... more Inthis paper, wepropose asystem whichdecides camera workina3Dvirtual soccer gameandgenerates 3DCGanimation automatically, called DataAnimating. The3Dvirtual soccer gameisrepresented bytheplayevents including player IDs, player positions, types ofplayandsoon.These properties arecreated byreal soccer video analysis system, called Digital Scorebook(1). Thesystem generates a3DCGanimation according totwouserselections ofsoccer playevents. Tocreate impressive soccer animation fromreal soccer game,wefocus onvirtual cinematography asthe artofrhetoric ofanimation scene. Thesystem canautomate acomposition ofanimation story anddecide camerawork based onthevirtual cinematography. Ouranimation story hasthethree part, Introduction, Development andConclusion, selected fromtheorganization ofChinese poetry. Theappropriate cameraworkisdecided ineachpart. Moreover, the picture composition foreachpart inascreen iscalculated based ongolden ratio. Intheexperimental result, thesystem cangenerate 3DCGanima...

IEEJ Transactions on Electronics, Information and Systems, 2020

IEEJ Transactions on Electronics, Information and Systems, 2016

It is well known that points on a plane in 3D world are related to corresponding image points in ... more It is well known that points on a plane in 3D world are related to corresponding image points in a view of a moving camera by projective translation. Good image features have robust projectivity under any camera movements. In the standard performance evaluation of image processing, real captured images of a scene are used ordinarily. However, it is not enough to evaluate in detail because the variation of camera angle and distance to target objects are limited and the capturing cost is expensive. During the early stage of the image processing development, the basic performance measurement should be the most important in an easy way. We propose a projectivity diagnosis method to measure the performance of local descriptor base template matching between a template image and reference images which are created by deforming the template image. This template matching consists of a feature image point extraction and a local descriptor matching. The proposed method evaluates the positional accuracy of the extracted feature points and the matching with local descriptor. Four metrics are introduced to evaluate the projectivity of template matching. In the experiment, our proposed diagnosis method expose the projectivity of SIFT, SURF, and ORB. SIFT showed the better robustness than the others.

2018 16th International Conference on Frontiers in Handwriting Recognition (ICFHR), 2018

This paper presents a method which reduces uncertainty of a position and a direction of an autono... more This paper presents a method which reduces uncertainty of a position and a direction of an autonomous robot by observing environment with a camera. In the proposed method, the state of the robot is represented by a state vector obeying a probability distribution. The robot creates and renews an environment map by considering the information from the mounted camera. Moreover, the robot revises the probability distribution on its state by comparing the map and the information from the camera. This attempt at avoiding the inconsistency in the map reduces the uncertainty of the state of the robot. We present experimental results to show the possibility of this method.