The main objective of this research is in automating the initialization phase of the MEMS-IMZ/GPS data integration. The motivation for this study is the special case where the before mentioned sensors are worn on a body (e.g. of an athlete) and where the usually used static assumptions (i.e. zero velocity) are difficult to satisfy. Nevertheless, the proposed methodology is also applied on terrestrial vehicles with the aims of reducing the user interactions in the reconstruction of the trajectory from the recorded data. The proposed identification of dynamic versus (quasi) static periods is based on wavelet decomposition of the inertial measurements. After presenting the bases of the process using the Continuous Wavelet Transform (CWT), the automated software's architecture is presented together with experiences carried in different dynamic environments. The trajectories calculated with the automated initialization are compared to those benefiting from the manual selection of the initialization periods based on experience and external knowledge of the underlying motion. As the differences between both approaches are negligible the new method is validated.