This paper presents a novel approach for cartoon motion capture from traditionally hand drawn cartoon animations. Unlike previous methods based on skeletal model or key-shape representation, the motion of cartoon character is represented by global affine transformation and local non-affine deformation using thin-plate splines (TPS). The shape of cartoon character is represented by its contour. We directly match shapes between adjacent cartoon frames (or key frames), and retarget recovered motion to a new character, which will move like original one. The novelty of our approach lies in the ability to capture and retarget cartoon motion by shape matching without prior model. Results are presented to show its ability.