We have developed so far an automatic discrimination system of human sleep EEG stages based on a waveshape recognition method. These systems were able to detect discrete stages (stage MT, W, 1, 2, 3, 4, REM). However, they are not sufficient to extract much information in detail. Therefore, in order to extract more precise information for sleep stages, we have tried to analyze the sleep EEG in
... [Show full abstract] its time-frequency domain by employing a continuous wavelet transform based on Gabor wavelets. In this paper, a modified wavelet transform method is proposed. It has the feature that the damping coefficient of Gabor wavelets is adjusted by a sigmoid function. In the experiments, compared with the ordinary wavelet transform with some constant damping coefficients, it has been confirmed that the modified wavelet transform method can have the adequate frequency resolution for analyzing frequency components of sleep EEG.