Video Quality Evaluation in IP Networks

05/28/2012

Due to the growing demand for video applications and services, the need for accurate and effective video quality assessment metrics is increasing. This thesis proposes and evaluates no-reference objective quality metrics for encoded videos transmitted over a lossy channel. For this purpose, the metrics consider both the effects of H.264/AVC video encoding and packet losses over IP networks. All the developed algorithms predict the video perceived quality based on elements extracted from the bitstream and on the information about the packet losses, taken from the headers of the received packets. Five quality assessment metrics for H.264/AVC videos impaired by packet losses are proposed and evaluated: 1) a simple model that accounts for the packet loss ratio; 2) a model that considers the frame type where the correspondent losses occur; 3) a model that considers the frame type and the movement in the video under analysis; 4) a model that considers the frame type, the movement in the video and the dependencies between frames; 5) a model that considers the frame type and statistical metrics taken from the packet loss pattern. The fifth model provided the best results, with quality predictions well correlated with subjective assessment data.

Authors in the community:

• 

  M

  Miguel Filipe Chan Chin

  ist156797

Supervisors of this institution:

• 

  T

  Tomás Brandão

  ist409145

• 

  M

  Maria Paula Dos Santos Queluz

  ist12401

electrical-engineering-electronic-engineering-information-engineering - Electrical engineering, electronic engineering, information engineering

eng - English

Instituto Superior Técnico