Comparison of Content-Aware Motion Estimation methodsfor effective H.264 Coding for Video Streaming
Akshay K. Chiddarwar, Anu Sahni · 2015
With advancements in technology and continuous upgrade in digitalization, a lot of demand for high quality video has gained remarkable popularity. Streaming this type of videos requires a high bandwidth and on slow networks it usually ends up in high buffer time with poor video quality. New services in streaming industry like Video on Demand(VoD), Internet Television (IPTV) have gained momentum due to advantage of availability in any region at anytime. Mobility and ease of access is the most important aspect user finds while signing up for any service. Decoding of raw high quality video requires high processing power and memory on client side resulting in poor user experience when the Internet is unstable. n this paper, we compare and present results for Exhaustive search (esa) and Uneven Cross Multi Hexagon Grid Search (umh) motion estimation methods in H.264 video encoding. Transcoded output video is evaluated for Structural Similarity Index (SSIM), PSNR by maintaining a Constant Bitrate (CBR) to understand the difference in perceived video quality. Also, performance characteristics like computation time, data rate is evaluated to understand the efficiency of these methods.