The DWT is defined by the following equation:
(t) = 2
(t-b)) (1)where a is a scales and b is positions of the wavelet
mother ψ (
) is a time function with finite energy. Choosingscales and positions are based on powers of two, which arecalled dyadic scales and positions (a
k) ( j
integers). Equation (1) shows that it is possible to build a
wavelet for any function by dilating a function ψ (
) with acoefficient 2 j, and translating the resulting function on a
grid whose interval is proportional to 2−j
The selection of appropriate wavelet and the number of decomposition levels is very important in analysis of signals using the WT. The number of decomposition levelsis chosen based on the dominant frequency components of the signal. The levels are chosen such that those parts of thesignal that correlate well with the frequencies required forclassification of the signal are retained in the waveletcoefficients. The smoothing feature of the Daubechieswavelet of order 2 (db2) made it more suitable to detectchanges of the signals. Thus, the wavelet coefficients arecomputed using db2. The frequency bands correspondingto different levels of decomposition for db2 with asampling frequency of 256 Hz. The discrete waveletcoefficients are computed using the MATLAB wavelettoolbox.The purpose of feature extraction is to reduce the sizeof the original dataset by measuring certain properties orfeatures that distinguish one input pattern from another.The various measurements based on statistical featuresfrom EEG are extracted. The extracted features provide thecharacteristics of the input type to the classifier byconsidering the description of the relevant properties of thesignals into a feature space.The statistical feature of the wavelet coefficients ineach subband such as energy, entropy, Minimum subband,maximum subband, mean, and standard deviation are usedto investigate the adequacy for the discrimination of normaland abnormal patients. The following statistical featureshave been derived using the following.
the diminished capacity for spontaneouschanges in signals.Entropy =
(2)Where P(i, j)
reflects the distribution of the probabilityof occurrence of each signal (i , j are integer).
Provides the sum of squared elements in thewavelet. This is also known as uniformity or the angularsecond moment.The energy is computed using E is given by
is signal value, values are present in waves isdenoted as n. Total number of signal is N
It generate maximum of thewavelet coefficients in each subband is calculated usingM
) (4)where max (xi) is maximum number of signal value.
It is defined as average value of a distributionof the wavelet coefficients in each subband which is givenby
where xi is signal and total number of signal is presentin the wavelet is N
calculate minimum of thewavelet coefficients in each subband is defined asM
) (6)Where min (xi) is minimum number of signal value.
- standard deviation of eachsubb
and is defined as σ .This feature provide information
about the amount of change of the frequency distribution.
where ∑ is sum of squared elements in the wavelet,x is
signal value and µ is a mean of the correspondingsignal(xi).Thus a total of 21 statistical feature are extracted fromEEG signal for each subband for preparing dataset.
(IJCSIS) International Journal of Computer Science and Information Security,Vol. 8, No. 6, September 2010118http://sites.google.com/site/ijcsis/ISSN 1947-5500