A major challenge in morphometrics is to analyse complex biological shapes formed by structures at different scales. Leaves exemplify this challenge as they combine differences in their overall shape with smaller shape variations at their margin leading to lobes or teeth. Current methods based on contour or on landmarks analysis are successful in quantifying either overall leaf shape or leaf margin dissection, but fail in combining the two. Here, we present a comprehensive strategy and its associated freely available platform for the quantitative, multiscale analysis of the morphology of leaves with different architectures. For this, biologically relevant landmarks are automatically extracted and hierarchized, and used to guide the reconstruction of accurate average contours that properly represent both global and local features. Using this method we established a quantitative framework of the developmental trajectory of Arabidopsis leaves of different ranks and retraced the origin of leaf heteroblasty. When applied to different mutant forms our method can contribute to a better comprehension of gene function as we show here for the role of CUC2 during Arabidopsis leaf serration. Finally, we illustrated the wider applicability of our tool by analysing hand morphometrics.