Wavelet shrinkage of itch response data

This article addresses the problem of denoising piecewise constant functions by using both an jump-tolerant moving average and Haar wavelet shrinkage. The piecewise constant functions are contaminated with Poissonlike noise and are measurements of perceived itch by human subjects in an experiment to relate perceived itch to blood ow in response to histamine iontophoresis. We show that the translation-invariant wavelet transform with universal wavelet shrinkage provides a fast and automatic method for producing good estimates of the underlying perceived itch and that the method is superior to the jump-tolerant moving average and standard wavelet shrinkage. The article then looks in more detail at the noise structure of the signals and highlights some new wavelet techniques that might