Comparison of filter banks of DWT in recovery of image using one dimensional signal vector

Abstract

Digital filter bank of DWT (discrete wavelet transform) provides the approximate and detail components of a signal, whereas the perfection of the filter bank depends on recovery of the signal. In this paper, we convert an image into a one dimensional signal vector then the perfection of the filter bank is observed on recovery of signal vector hence the image. Here orthogonal FIR (finite impulse response), FIR under window method, IIR (infinite impulse response) Butterworth, Chebyshev type-I, type-II and synchrosqueezed transform are compared in context of image recovery under statistical parameters: mean error, variance of error, difference of skewness and kurtosis, cross-correlation coefficient, PSNR (peak signal to noise ratio) and histogram. The orthogonal FIR combination gives the best results and synchrosqueezed transform is the worst case for reconstruction of image.