Kousuke Kumamaru

FEATURE EXTRACTION OF HUMAN SLEEP EEG BASED ON A PEAK FREQUENCY ANALYSIS

Abstract We have developed so far the automatic discrimination system of human sleep EEG stages based on a wave-shape recognition method. These systems were able to detect discrete stages (Stage MT, W, 1, 2, 3, 4, REM). But, more detailed information extraction was impossible by them. Therefore, in this paper, continuous wavelet analysis is applied to EEG signals in order to extract more precise information for the stages. A modified wavelet transform method is proposed and an extraction method of time series of peak frequency based on time-frequency analysis is introduced. And it is confirmed that our method is effective through the experimental studies.

Pattern recognition of EEG signals during right and left motor imagery

In this paper, statistical pattern recognition method based on AR model was introduced to discriminate the electroencephalograph (EEG) signals recorded during right and left motor imagery. And learning methods were investigated. Also, correlation between C3 and C4 signals were investigated, and thereby which AR (combine AR or multivariable AR) model must be used in each EEC recording method.

Fault detection of non-linear systems based on multi-form quasi-armax modeling and its application to the ship benchmark

Fault detection of nonlinear systems based on multi-form quasi-ARMAX modeling and its application to the ship benchmark

IMPROVEMENT OF FAULT DETECTION METHOD FOR NONLINEAR BLACK-BOX SYSTEMS BASED ON MULTI-FORM QUASI-ARMAX MODELING

Improvement of fault detection method for nonlinear black-box systems based on multi-form quasi-ARMAX modeling

Segmentation of range data based on a stochastic clustering method with competitive process

In this paper, a stochastic clustering method with a competitive process is proposed to segment significantly the entire circumferential range data. The segmentation technique is utilized as the preprocessing of 3-D shape modeling so that the modeling can be more easily achieved for the object that has arbitrary topology, in which the data points are divided into the several subsets that represent the 3-D shapes of different quadric surfaces. The clustering method is implemented by evaluating a distance computed between each data point and each quadric surface. Furthermore, it consists of creation and competitive processes in order to obtain the desirable clusters. Consequently, since the only appropriate clusters are remaining, the segmentation can be achieved by assigning the data points to these clusters.