Three-dimensional (3D) dynamic mesh sequences, representing a character/animal, an object or an organ evolving through time, are nowadays more and more used in diverse applications such as digital entertainment, computer-aided design and medical imaging. It is common that a dynamic mesh undergoes some lossy operations, e.g. compression, simplification and watermarking. In most cases the end users of dynamic meshes are human beings; therefore, it is important to evaluate how much visual distortion has been introduced by a certain lossy operation and whether this distortion deteriorates the quality of service related to the mesh sequence. In this thesis project, we are interested in the objective perceptual quality assessment (i.e. by means of well designed software tools, as opposed to subjective assessment conducted by human observers) of 3D dynamic meshes and its applications. All existing objective quality metrics, either for static or dynamic meshes, are based on “top-down” assumptions about human visual system (HVS) that are rather difficult to verify. In this project, we will adopt a different “bottom-up” approach which is based on the experimental study of HVS properties when humans observe dynamic meshes. The objectives of the proposed thesis are threefold: (1) First, we will try to better understand HVS properties when human subjects observe 3D mesh sequences, through a series of psychophysical experiments. In particular, we are interested in the contrast sensitivity of HVS to different spatiotemporal components of dynamic meshes and the visual masking effect of HVS when observing animated meshes. (2) Second, based on the learned HVS features, we will attempt to devise an efficient and open-source “bottom-up” metric for the objective evaluation of the perceptual distance between two dynamic meshes. (3) Third, we want to explore the practical applications of the HVS features and the devised metric, especially in the case of dynamic mesh compression and/or watermarking.