Implementation of de-noise DWT chip based on adaptive soft-threshold

According to the relationship of wavelet transforms and perfect reconstructional FIR filter banks, this paper presents a real-time chip with adaptive Donoho's non-linear soft-threshold for de-noising in different levels of multi-scale space through rearranging the input data during convolving and filtering and sub-sampling, and more important, it gives a simple iterative algorithm to calculate the variance of the noise in interregna with no signal. It works well whether the signal or noise is stationary or not.