Presentation is loading. Please wait.

Presentation is loading. Please wait.

Wavelet-based Compression of 3D Mesh Sequences

Similar presentations


Presentation on theme: "Wavelet-based Compression of 3D Mesh Sequences"— Presentation transcript:

1 Wavelet-based Compression of 3D Mesh Sequences
Frédéric Payan Coauthor: Marc Antonini I3S laboratory - CReATIVe Research Group Universite de Nice Sophia Antipolis FRANCE 2nd ACIDCA-ICMI, Tozeur, Tunisia, november, 2005.

2 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.
Motivations Design a compression algorithm for mesh sequences (with a fixed connectivity). 23/02/2019 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.

3 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.
Problem Statement Raw representation: set of irregular meshes Talking Face: meshes of 539 vertices Each frame defined by the coordinates of its vertices VRML: more than 137 Mb! Compression needed… 23/02/2019 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.

4 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.
Summary Background Related works Contributions Proposed temporal lifting scheme Bit allocation Simulation results Conclusion & perspectives 23/02/2019 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.

5 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.
Summary Background Related works Contributions Proposed temporal lifting scheme Bit allocation Simulation results Conclusion & perspectives 23/02/2019 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.

6 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.
I. Background Main Related Works Principal component analysis (PCA): Alexa & Müller (2000) exploit the temporal correlation of sequence geometry but… very complex and long analysis step and efficient only when few shape deformations PCA+linear prediction coding (LPC): Karni & Gotsman (2004) more efficient than PCA but… same problem than PCA! Spatial multiresolution analysis: Guskov & Khodakovsky (2004) efficient to capture local shape deformations but… less efficient when global deformations 23/02/2019 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.

7 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.
I. Background Contributions Observation: no temporal wavelet-based coder ! Proposed approach: Temporal lifting scheme to exploit the temporal coherence fast and low complex synthesis but also analysis Bit allocation process optimal coding 23/02/2019 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.

8 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.
Summary Background Related works Contributions Proposed temporal lifting scheme Bit allocation Simulation results Conclusion & perspectives 23/02/2019 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.

9 Lifting scheme: principle
II. Proposed temporal lifting scheme Lifting scheme: principle Split -P U + HF details (wavelet coefficients) Low frequency (LF) signal Signal 3 steps: split => two cosets Prediction operator => set of details Update operator => low frequency (LF) signal defined by a pair [n,m] 23/02/2019 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.

10 Temporal Lifting scheme: principle
II. Proposed temporal lifting scheme Temporal Lifting scheme: principle Time Original Sequence TLS LF Sequence Sets of HF details 23/02/2019 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.

11 Temporal Lifting Scheme for mesh sequences
II. Proposed temporal lifting scheme Temporal Lifting Scheme for mesh sequences Principle: for each vertex, apply a 1D lifting scheme on its successive positions along the time axis 23/02/2019 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.

12 Temporal Lifting Scheme for mesh sequences
II. Proposed temporal lifting scheme Temporal Lifting Scheme for mesh sequences Example: filter [2,0], with 2 levels of decomposition Successive positions of the vertex V(i) wavelet coefficients LF data wavelet coefficients LF signal 23/02/2019 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.

13 Temporal Lifting Scheme for mesh sequences
II. Proposed temporal lifting scheme Temporal Lifting Scheme for mesh sequences => Multiresolution decomposition Input sequence time First set of temporal details h(1)(1) h(1)(2) h(1)(3) h(1)(4) Second set of temporal details h(2)(1) h(2)(2) LF sequence Decomposition level 23/02/2019 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.

14 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.
Summary Background Related works Contributions Proposed temporal lifting scheme Bit allocation Simulation results Conclusion & perspectives 23/02/2019 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.

15 Bit allocation: principle
III. Bit allocation Bit allocation: principle D Compression: optimize the rate-distortion tradeoff Multiresolution representation => how dispatching the bits across the different sets of details ? R Solution: use a bit allocation to maximize the quality for a user-given bitrate Decomposition level h(1)(1) h(1)(2) h(1)(3) h(1)(4) h(2)(1) h(2)(2) Bit budget ? 23/02/2019 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.

16 Proposed bit allocation
III. Bit allocation Proposed bit allocation Objective: find the quantization steps (used to encode the different set of details) that minimize the MSE of the reconstructed mesh sequence for a user-given target bitrate. This problem can be modeled by 23/02/2019 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.

17 How solving this problem?
III. Bit allocation How solving this problem? Find the quantization steps and lambda that minimize the following lagrangian criterion: Method: first order conditions Algorithm: iterative and model-based Distortion Constraint relative to the bitrate 23/02/2019 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.

18 connectivity encoding
III. Bit allocation Overall coding scheme connectivity encoding Sequence connectivity MUX 1011… Temporal DWT SQ Entropy Coding Geometry encoding {q} Target Bitrate Bit Allocation SQ: scalar quantization Connectivity encoding: valence-based coder of Touma & Gotsman 23/02/2019 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.

19 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.
Summary Background Related works Contributions Proposed temporal lifting scheme Bit allocation Simulation results Conclusion & perspectives 23/02/2019 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.

20 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.
III. Simulation results Distortion measure K&G in [2] => relative discrete L2-norm both in time and space with: matrix G: geometry of the sequence matrix hat G: quantized geometry Matrix E(G): mean coordinates of each frame 23/02/2019 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.

21 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.
III. Simulation results Different models! Talking face Few global motion Surprised chicken high global motion high local deformations Mad cow few local deformations 23/02/2019 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.

22 Different lifting schemes…
III. Simulation results Different lifting schemes… Filter [4,2]: less computing resources in time and memory 23/02/2019 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.

23 Different numbers of decomposition…
III. Simulation results Different numbers of decomposition… high motion => few decomposition levels (4) few motion => a lot of decomposition levels (7) 23/02/2019 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.

24 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.
III. Simulation results Comparison with… TG (Touma & Gotsman) Dynapack (Ibarria & Rossignac) PCA (Alexis & Müller) KG (PCA+LPC, Karni & Gotsman) WCA (Guskov & Khodakovsky) on… 23/02/2019 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.

25 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.
III. Simulation results …The talking face! few global motion 23/02/2019 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.

26 …the surprised chicken!
III. Simulation results …the surprised chicken! high global motion and local deformations 23/02/2019 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.

27 …the mad cow! high global motion and few local deformations
III. Simulation results …the mad cow! high global motion and few local deformations 23/02/2019 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.

28 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.
Summary Background Related works Contributions Proposed temporal lifting scheme Bit allocation Simulation results Conclusion & perspectives 23/02/2019 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.

29 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.
IV. Conclusions and perspectives Conclusions Temporal lifting scheme for mesh sequences simpler and faster than other analysis tools (for analysis and synthesis) like PCA… Optimal encoding of the different subbands Model-based bit allocation => Efficient and fast compression method for mesh sequences 23/02/2019 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.

30 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.
IV. Conclusions and perspectives Perspectives Additional spatial multiresolution analysis on LF sequences Explicit motion estimation technique (like in video coding) ? should reduce geometry information transmit side information (motion vectors) 23/02/2019 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.

31 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.
This is the end… My fpayan.i3s.unice.fr My webpage: Acknowledgement Microsoft Inc. For the surprised chicken D. Terzopoulos For the talking face Z. Karni for providing us with these data I. Guskov for providing us with mad cow Results extracted from paper of Z. Karni & C. Gotsman provided by I. Guskov 23/02/2019 2nd ACIDCA-ICMI, Tozeur, Tunisia, november 2005.


Download ppt "Wavelet-based Compression of 3D Mesh Sequences"

Similar presentations


Ads by Google