Motion vector estimation using parallel processing
Suvojit Acharjee, Gautam Pal, Rédha Taïar, Sayan Chakraborty, Sheli Sinha Chaudhuri, Nilanjan Dey · 2014
Video compression is the most active research area in multimedia and signal processing. The most important component in multimedia is video. From surveillance to disease diagnosis video has numerous applications. Videos generally contain huge temporal redundancy. Hence, such video requires the use of motion vector estimation technique in order to remove the temporal redundancy. Although motion vector estimation is computationally expensive. Different algorithms has been developed to reduce the computational complexity of motion vector calculation. All of these algorithms were serial. In this paper, motion vector is calculated by processing multiple frames parallely. Block based exhaustive search is used for calculation of motion vector. Computation result shows, this approach speeds up the process by 40% and having the same quality of recovered frame compared with the quality of recovered frames in serial processing.