Wave propagation in soft and hard biological tissues (such as bones) generates ultrasound signals forming packages with different time and frequency signatures. Correlating the complex physical processes involved with these packages means separating the useful information from parasitic one (e.g. electronic noise). This study proposes a way to increase the ultrasound signals resolution, aiming at simultaneously analyzing the time and frequency parameters, by using a wavelet decomposition of the signals. We also discuss ways to generate suitable transmitted signals correlated with the parameters of the test system. This deconvolution method based on wavelets enables to directly determine the transfer function of the scatterer, and gives improved results compared to more classical schemes.