A multi-scale approach to investigate liquid atomization processes is introduced. It describes the liquid system by the scale distribution whose determination is inspired from the Euclidean Distance Mapping used to measure the fractal dimension of a contour. The scale distribution is introduced in 2D and in 3D and applications from previous investigations are presented. The 2D applications are performed on experimental images and the 3D applications are performed on results obtained from Direct Numerical Simulation. The multi-scale analysis allows identifying and quantifying the mechanisms responsible for the interface evolution according to the scale. Among other results, the analyses presented here demonstrate the improvement of the atomization process when an elongation mechanism contributes to the thinning of the small structures. The multi-scale tool also provides new metrics that may be used to validate simulation results. An example of this is presented and discussed. Finally, the paper evokes several approaches to implementing the scale-distribution concept to improve or build new models.