Optimal wavelet packet decomposition for rectal pressure signal feature extraction

Abstract

The method of optimal wavelet packet decomposition is proposed for rectal pressure signal feature extraction. By using wavelet packet algorithm, the mean wavelet coefficients and its corresponding energy component with high separability are selected as the feature vector according to the maximum separation degree of Fisher index, and the optimal features vector have specific sub-band wavelet packet coefficients and energy with higher separability. By comparison of the classification result and the operation time of optimized and non-optimized features vectors, the experimental results give the evidence that the proposed method is effective.