Volume 18, No. 6, 2021
Video Saliency Detection Using Modified Hevc And Background Modelling
Mrs. Sharada P N , Dr. S C Lingareddy
Abstract
The world is now heavily dependent on live streaming and video conferences with the advent of the Covid – 19 pandemic. This change has led to the formation of several algorithms that specialise in video saliency. Our proposed algorithm is a modified HEVC algorithm that employs background modelling and implication of classification into coding blocks. With the help of G-Picture in the fourth long term reference position and usage of coding blocks, the overall coding complexity along with time consumption has been reduced as well as the efficiency of compression has increased. The algorithm is tested on the DHF1K dataset and has been compared with several state-of-the-art methods and the final result showcases that the proposed solution has the best overall compression efficiency and accuracy.
Pages: 1322-1330
Keywords: The world has been trying to copy the process of the human eye of filtering the unnecessary parts of what we are viewing and only considering the important ones. Image saliency has been a well-researched field while video saliency has not been so.