1 / 19

Fast motion estimation and mode decision for H.264 video coding in packet loss environment

Presenter : SeungDae Jeong. Fast motion estimation and mode decision for H.264 video coding in packet loss environment. Li Liu, Xinhua Zhuang Computer Science Department, University of Missouri-Columbia, USA IEEE ICME(International Conference on Multimedia & Expo) 2008 June 2008.

magee
Download Presentation

Fast motion estimation and mode decision for H.264 video coding in packet loss environment

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Presenter : SeungDae Jeong Fast motion estimation and mode decisionfor H.264 video coding in packet loss environment Li Liu, Xinhua Zhuang Computer Science Department, University of Missouri-Columbia, USA IEEE ICME(International Conference on Multimedia & Expo) 2008 June 2008

  2. Introduction • Packet loss environment • Practical video communications • Videos are usually coded without considering varying channel conditions • Error propagation

  3. Introduction Intra frame Previous frame Inter frame with packet loss • “Intra refresh” algorithm • Using intra frames that doesn't refer to previous frames • The decision of intra mode • Rate distortion optimization • Quantization distortion • Channel distortion • Simulation based estimation • Model based estimation

  4. Contents • RD Optimized intra fresh for error resilience • Lagrange Cost Function in Packet Loss Environment • Model Based Channel Distortion Estimation • Use channel distortion map for fast mode decision and motion estimation • Channel distortion map based fast mode decision and motion estimation algorithm • Fast Mode Decision • Fast Motion Estimation • Experiment results • Conclusion

  5. Lagrange Cost Function in Packet Loss Environment The source distortion by quantization The result bit rate The relation between distortion and bit rate Set of all defined coding options for macroblock m in frame n

  6. Lagrange Cost Function in Packet Loss Environment The estimated channel packet loss rate The quantization distortion The error propagation distortion if reference frames in motion estimation are erroneous The error concealment distortion when a macrobloak is lost In packet loss environment

  7. Lagrange Cost Function in Packet Loss Environment The modified Lagrange cost function for RD optimization mode decision The rate distortion cost for intra mode

  8. Model Based Channel Distortion Estimation • Define a channel distortion map on 4x4 block basis • To track potential error propagation

  9. Model Based Channel Distortion Estimation The optimal mode decision The error propagation distortion of ith block The error concealment distortion The estimated error propagation distortion Block based channel distortion map

  10. Use channel distortion map for fast mode decision and motion estimation • The error propagation distortion tends to surpass the quantization distortion as • frames contain more motions (e.g., from frame 150 to 250) • loss rate increases (e.g., from 0.05 to 0.10)

  11. Use channel distortion map for fast mode decision and motion estimation • Statistical information from coded video sequence using original MBDE algorithm

  12. Channel distortion map basedfast mode decision and motion estimation algorithm • Fast mode decision • Computational cost of estimating Dep(o) is quite small • By using Dep(o) as assistance to decide early termination of possible inter coding choices, encoding speed can be accelerated significantly • Fast Motion Estimation • Choosing the best motion vector • Minimizing a Lagrange cost function

  13. Fast mode decision • Computational cost • Inter coding cost > Intra coding cost • The calculation of minDep • Dependent on optimal motion vector decision from motion estimation process • Computational cost can not be reduced • Computing Dep(o) based on the predicted motion vector

  14. Fast mode decision Computing Dep(o) based on the predicted motion vector

  15. Fast motion estimation The sum of absolute difference The number of bits to code motion information • Minimizing a Lagrange cost function • Fast motion estimation algorithm selects best motion vector from a set of candidates, and terminates those macroblocks that may have large error propagation distortion associated

  16. Fast motion estimation

  17. Experiment results • Implemented using H.264 reference software JM12.1 • http://iphome.hhi.de/suehting/tml/download/old_jm/ • Video sequences • Carphone, Foreman, Coastguard • QCIF format, 7.5 fps • Fixed quantization parameter (QP) • Packet loss simulation • S. Wenger, “Internet Error Patterns”

  18. Experiment results • PSNR • Peak Signal-to-Noise Ratio • Bit rate • Speed • Very fast

  19. Conclusion Two stage fast mode decision and motion estimation algorithm for H.264 video coding in packet loss environment Advantage of estimated channel distortion as assistance to decide early termination of possible inter coding options Reduce coding complexity significantly while maintaining similar average picture quality

More Related