Advanced Multimedia Processing Lab -- Group Member -- Deepak Turaga

            About AMP Lab        Projects        Downloads        Publications        People        Links

Group Member

Deepak S. Turaga
PhD. Alumnus
Deepak is currently a Research Staff Member in the Content Server Department at
IBM T. J. Watson Research Center in Hawthorne, NY.

Email: turaga(at)us(dot)ibm(dot)com

 

Phone: 914-784-7550
Fax: 914-784-6226

Mailing Address:
IBM T. J. Watson Research Center,

19 Skyline Drive,

Hawthorne, NY 10532

[Research Interests]        [Project]      [Publications

Research Interests

         Current research interests

o        Multimedia streaming

Content and Server Management, Multiple Description Coding, Error concealment, Cross-layer optimization for increased error resilience

o        Interframe wavelet video coding

Motion compensated temporal filtering and various improvements to improve the flexibility and efficiency of wavelet video coding

         Research work performed for my Ph. D. thesis

o        Encoder Optimization Issues for Video Coding

         Block based motion estimation  is an integral part of video coding standards like H.263 and MPEG 1,2 and 4. We can use the spatial correlatedness of motion in video sequences to reduce the search space, provide good matches and also reduce the bit rate (as motion vectors are differentially coded, a homogeneous motion field needs fewer bits to code). The performance of our algorithms is evaluated in terms of this "speed-quality -bit rate" tradeoff.

         New measure of similarity  and mode decisions for Intra versus Inter coding. We propose a new measure of similarity between blocks called the mean reduced MAD, which is more representative of the number of bits required for coding the residue as opposed to the MAD. Using this measure can lead to better performance in terms of motion estimation quality. This measure can also be used to provide a better mode decision for Intra versus Inter coding.

         Classification Based Mode Decisions. Mode decisions in the video coding process may be viewed as binary hypothesis testing problems and a suitable classifier may be designed to make these decisions. We choose representative features for the hypotheses under test and design classifiers to minimize the overall cost of misclassification. This cost may be defined in terms of the speed-quality-bit rate parameters of video coding. We may use these classification based strategies to make any mode decisions, such as the Inter-Intra mode decision for a macroblock or the I-P frame decision or the code frame, skip frame selection for rate control. 

o        Network Adaptive Video Coding

         Video source trace modeling. We propose some doubly stochastic models (using HMMs) to generate video bit rate traces. These traces can be used for estimating network parameters such as delay, bit error rates without the need for actual video codecs. The models are designed to mimic the statistical characteristics of real video trace data as closely as possible. 

         Rate Control using different kinds of scalabilities provided by video coding standards adaptively. We may use the classification based rate control strategies to choose between coding a frame and skipping it for rate control and optimize the decision in terms of the rate-distortion parameters.

         Error Concealment  for video transmitted over lossy networks. Current work is on model based error concealment, using Principal Components Analysis as a model for the video data region of interest. 

o        Computer Vision

         Gesture recognition  using HMMs. Some work that I did as part of my internship at Intel during Summer 1999.

Top of this page

 

Project

H.263 Video Codec

For more details on the project please see the Video Codec project page.

Top of this page

                                  

Publications

  • Book Chapters
  • Deepak S. Turaga and Tsuhan Chen, Book Chapter "Fundamentals of Video Coding,: H.263 as an example",  Compressed Video Over Networks, The Signal Processing Series, Marcel Dekker, 2000.
  • Deepak. S. Turaga and M. van der Schaar, “Tutorial on MPEG-4” under preparation.
  • Journal Papers
  • Deepak S. Turaga, M. van der Schaar et al, "Unconstrained motion compensated temporal filtering for efficient and flexible interframe wavelet video coding," accepted for publication, Signal Processing: Image Communication.
  • Deepak S. Turaga, M. van der Schaar and B. Pesquet-Popescu, “Complexity scalable motion compensated wavelet video encoding,” under review IEEE Trans. CSVT.
  • D. S. Turaga, Y. Chen and J. Caviedes, “No reference PSNR estimation for compressed pictures,” Signal Processing: Image Communication, Special Issue on Video Quality, Jan 2004.
  • Deepak S. Turaga and Tsuhan Chen, "Model-based error concealment for wireless video," IEEE Trans. CSVT, Special Issue on Wireless Video, 2002.
  • Deepak S. Turaga and Tsuhan Chen, "On probability density for modeling video traffic," Journal of VLSI Signal Processing, 2002.
  • Deepak S. Turaga and Tsuhan Chen, "Classification based mode decisions for video over networks," IEEE Trans. Multimedia, Special issue on Multimedia over IP, 2001.
  • Deepak S. Turaga and Tsuhan Chen, "Estimation and Mode Decision for Spatially Correlated Motion Sequences", IEEE Trans. CSVT, May 2002.
  • Conference Papers
  • M van der Schaar, D. S. Turaga, V. Akella, “Rate-distortion-complexity adaptive video compression and streaming,” ICIP 2004.
  • D. S. Turaga and M. van der Schaar, “Temporal prediction and differential coding of motion vectors in the MCTF framework,” ICIP, September 2003.
  • M. van der Schaar and D. S. Turaga, “Multiple description scalable coding using wavelet-based motion compensated temporal filtering,” ICIP, September 2003.
  • D. S. Turaga and M. van der Schaar, “Reduced complexity spatio-temporal scalable motion compensated wavelet video encoding,” ICME, July 2003.
  • Deepak S. Turaga and M. van der Schaar, "Content-adaptive filtering in the UMCTF framework", ICASSP 2003
  • M. van der Schaar and Deepak. S. Turaga, "Unconstrained motion compensated temporal filtering (UMCTF) framework for wavelet video coding," ICASSP 2003.
  • Deepak S. Turaga, M. van der Schaar and B. Pesquet-Popescu, "Differential Motion Vector Coding for Scalable Coding," SPIE Electronic Imaging 2003.
  • Deepak S. Turaga and M. van der Schaar, "Wavelet coding for video streaming using new unconstrained motion compensated temporal filtering," International Workshop on Data Communication 2002.
  • Deepak S. Turaga and M. van der Schaar, "Multiple description motion compensated temporal filtering for robust wireless video transmission," 2nd New York Metro Area Networking Workshop, August 2002.
  • Deepak S. Turaga and M. van der Schaar, "Wavelet coding for video streaming using new unconstrained motion compensated temporal filtering," International Workshop on Data Communication 2002.
  • Deepak S. Turaga, Y. Chen and J. Caviedes, "No reference PSNR estimation for compressed pictures," ICIP, Rochester, Sept. 2002.
  • Deepak S. Turaga and T. Chen, "Face recognition using mixtures of principal components," ICIP, Rochester, Sept., 2002.
  • Deepak S. Turaga and Tsuhan Chen, "I/P frame selection using classification based mode decisions," ICIP 2001.
  • Deepak S. Turaga and Tsuhan Chen, "Hierarchical Modeling of Variable Bit Rate Video Sources," Packet Video 2001.
  • Deepak S. Turaga and Tsuhan Chen," Activity-adaptive modeling of dynamic multimedia traffic," ICME 2000.
  • Deepak S. Turaga and Tsuhan Chen, "Modeling of Dynamic Video Traffic," ISCAS 2000
  • Mohamed Alkanhal, Deepak Turaga and Tsuhan Chen, "Correlation Based Search Algorithms for Motion Estimation", Picture Coding Symposium, April 1999
  • David Casasent, Wei Su and Deepak Turaga, "SAR ship detection using new conditional contrast box filter", SPIE’s 13th Annual International Symposium on Aerospace/Defense Sensing, Simulation and Controls, April 1999.

 

  • Theses
  • Ph.D.Thesis "Statistical Modeling for Networked Video: Coding Optimization, Error Concealment and Traffic Modeling", July 2001
  • M.S. Thesis, “Encoder Optimizations for Networked Video,” May 1999.

Top of this page


This website is maintained by Devi Parikh
Copyright 2001 Advanced Multimedia Processing Lab. All rights reserved.
Revised: July 15, 2006.