The analysis of EEG signals plays a vital role in the detection of Seizure. The EEG signal of a normal person varies when compared to that of a seizure affected person. A new wavelet is created which closely represents a normal EEG wave. The discrete wavelet transform using the new wavelet family is applied to the input EEG signals. Since the new wavelet represents the normal EEG wave pattern the reconstruction error is very less when a normal EEG signal is applied as an input when compared to that of a seizure EEG signal. This marked difference in the reconstruction error is used to classify the normal and seizure EEG signals. When continuous wavelet transform is applied to the input EEG signal using the new mother wavelet, the wavelet coefficients are higher in seizure EEG when compared to that of the normal EEG signal.