Paper Title
Parallelizing And Comparing Block Matching Algorithms In H.264 On Nvidia Cuda

Judging by the amount of research work that is being carried out in the field of parallel computing, it seems to be the next big thing and it is safe to say that this phenomenon is here to stay. H.264 is a powerful yet flexible video encoding standard which provides better video quality at lower bitrates. The motion estimation part of video compression is very time consuming. This computational time can be reduced by parallelizing it on NVIDIA’s CUDA a GPGPU which ensures a better performance than its non-parallel counterpart using only CPU. This paper strives to provide a comprehensive look into the three block matching search algorithms, namely Full Search, Three Step Search and Uneven Multi Hexagon Search used in motion estimation. We parallelize and implement the algorithms in H.264 on CUDA and compare their results to decipher which provides better efficacy.