ROI based video compression tailored for video communication.

In this project , Region of Interest (ROI) based resource allocation is done in Quantization stage of Video Compression to achieve psychovisual compression in video communication scenario. The Quantization depends heavily on the transform stage. The motion compensation stage depends heavily on both Transform and Quantization stage. So several experiments are done at different levels to see how ROI priority parameters affect each of these stages. For detecting ROI Viola Jones face detector is used. For the transform stage Integer transform and Discrete Cosine transform are used in parallel and their behavior along with ROI priority parameters is studied.

Presentation - Download

Report - Download

Matlab Code - Github