I've been playing around with multimodal probability distributions lately. As you probably know there exist extremely many methods to try and identify the modes / sub-distributions in probability distributions.
Here is a rather nice and simple example showing the histogram of three gaussians of different $\mu$ and $\sigma$. We want to determine how many there are and calculate the mean value and standard deviations of each.
However I am not aware of any methods which use wavelet analysis although my intuition tells me there should exist plenty because wavelets can be designed to be rather good at multiscale peak detection. Any reference to papers or other sources which handle this would be welcome, (or sketch of any own approach also).