Neighbour Block Difference Vector (Nbdv) Based Motion Estimation and Self Occlusion Detection in Video Compression
K. Vidyavathi, R. S. Sabeenian · Transylvanian Review · 2016
In this work, a neighboring block-based difference vector (NBDV) method is proposed. The basic concept of NBDV is to derive a difference between current frame and predicted future frame motion information by using spatial and temporal vectors. This work presents a novel approach to extract reliable dense and distance motion estimation for human gesture in a video sequence. A difference vector by definition is not a motion vector and is not directly used for motion compensation, but used for advanced inter-view predictions.A variation on the moving vector is a motion vector concerning index indicating an inter-view reference picture. A temporal motion vector is a motion vector with reference index indicating a reference image from the same view. Difference vector is a vector used to connect relevant blocks of two views and different from a disparity motion vector, is not transmitted as part of the bitstream. Then the self-occlusion is detected by an interconnected vector-based method and an adaptive motion estimation method between successive frames. An experiment on a test video sequences shows that the proposed method guarantees the integrity and accuracy of dense points are tracking and performs better than several state-of-the-art methods. The proposed method is suitable for applications where the accuracy of the motion estimation is vital.