Current Projects

Fast Mode Selection and Motion Estimation for H.264 Encoders Key Words: Low-complexity algorithms, video encoding, mode selection

Project Members

• Osman G. Sezer

Project Description

Coding mode selection is one of the most important modules in digital video encoding. Video compression standards provide multiple strategies (modes) in order to encode a source block image efficiently. In general, mode selection should be based on rate constraints or/and distortion requirements, etc. Lagrangian rate-distortion optimization which simulates actual encoding process is successfully applied to mode selection problem. However, it is almost impossible to implement the method in most power-limited or computation-limited applications because of its prohibitive computational complexity. Our research efforts focus on low complexity mode selection method while maintaining rate-distortion optimality. Another important encoder module is motion estimation. Motion estimation is the most computationally intensive module in video encoders. Especially, in the latest video compression standard, H.264/AVC, variable motion compensation blocks are introduced, which means that each block requires motion estimation. We focus on reducing adaptively the number of motion compensation blocks as well as developing fast motion estimation algorithms.

Related Publications

• H. Kim, N. Kamaci, and Y. Altunbasak, “Low-complexity rate-distortion optimal macroblock mode selection and motion estimation for MPEG-like video coders,” IEEE Transactions on Circuits and Systems for Video Technology, Volume 15, Issue 7, July 2005 Page(s):823 - 834 [pdf]
  (The IP in this paper is patented, licensed and incorporated into a product. For more information, contact yucel@ece.gatech.edu)
• H. Kim and Y. Altunbasak, “Low-complexity macroblock mode selection for the H.264/AVC encoders,” IEEE Int. Conf. on Image Processing, Suntec City, Singapore, October 2004 [pdf]
• H. Kim and Y. Altunbasak, “Rate-distortion optimal joint macroblock mode selection and motion estimation for MPEG-like video coders,” IEEE Int. Conf. on Image Processing, Suntec City, Singapore, October 2004 [pdf]
• H. Kim and Y. Altunbasak, “Low-complexity rate-distortion optimal macroblock mode selection for MPEG-like video coders,” in IEEE Int. Conf. Multimedia and Expo, Baltimore, MD, July 2003 [pdf]