Compressing video based on region of interest

Real-time video streaming suffer from bandwidth limitation that are unable to handle the high amount of video data. To reduce the amount of data to be streamed, we propose an adaptive technique to crop the important part of the video frames, and drop the part that are outside the important part; this part is called the Region of Interest (ROI). The Sum of Absolute Differences (SAD) is computed to the consecutive video frames on the server side to identify and extract the ROI. The ROI are extracted from the frames that are between reference frames based on three scenarios. The scenarios been designed to position the reference frames in the video frames sequence. Linear interpolation is performed from the reference frames to reconstruct the part that are outside the ROI on the mobile side. We evaluate the proposed approach for the three scenarios by looking at the size of the compressed videos and measure the quality of the videos by using the Mean Opinion Score (MOS). The results show that our technique significantly reduces the amount of data to be streamed over wireless networks with acceptable video quality are provided to the mobile viewers.