| Wavelet analysis is a new technique in time and frequency analysis domain recent years. It is the inheritance and development of Fourier analysis, but it has much advancement than Fourier in essence. Because wavelet analysis is rigorous in theory, and it has the function of self-adaptive localization in time and frequency, it gets more used in signal procession domain and acquire much fruit. EEG map is currently a popular non-invasive brain function of advanced detection technology. It can help people better understand the mechanism of the brain activity, as well as people's cognitive processes and diagnose brain disease. Noise removal for EEG data processing is the primary aspect, this article apply wavelet technology to the neural information field, mainly do the treatment of EEG de-noising .At the last , it is compared by Fourier de-noising. The paper's main job is as follows:1. This article described in detail the physiological basis of EEG, acquisition methods, EEG characteristics, classification, principle and other related knowledge.2. I have studied the Fourier transform, wavelet transform and their properties, and proved their characteristics from math aspect in signal processing, demonstrated that the wavelet transform has more significant advantage than the traditional Fourier transform in non-stationary signal processing.3. Many kinds of signal de-noising methods are studied in this article. On the foundation of analyzing these methods' advantage and shortage, I select the threshold de-noising method, and it achieved good effect.4. The experiment's flow of EEG Acquisition- Data conversion- De-noising treatment-Result evaluation is implemented. The data from scan4.3 and matlab7.0 is unification. I successfully settled the problem. The original signal is de-noised by using Fourier and wavelet separately. I evaluated the de-noising result from SNR, RMSE, and energy ratio. Each indicator justifies that wavelet transform is better than Fourier transform in EEG signal de-noising. |
PDF Full Text Request