Segmentation-based mesh design for motion estimation

In most video compression standards today, the generally accepted method for temporal prediction is motion compensation using block matching algorithm (BMA). BMA represents the scene content evolution with 2-D rigid translational moving blocks. This kind of predictive scheme usually leads to distortions such as block artefacts especially when the motion is important. The two most important aims in video coding are to receive a good quality on one hand and a low bit-rate on the other. This thesis proposes a motion compensation scheme using segmentation-based 2-D triangular mesh design method. The mesh is constructed by irregularly spread nodal points selected along image contour. Based on this, the generated mesh is, to a great extent, image content based. Moreover, the nodes are selected with the same method on the encoder and decoder sides, so that the only information that has to be transmitted are their motion vectors, and thus very low bit-rate can be achieved. Compared with BMA, our approach could improve subjective and objective quality with much less motion information."--Résumé abrégé par UM