{ Fast Disparity Estimation Using Spatio- temporal Correlation of Disparity Field for Multiview Video Coding Wei Zhu, Xiang Tian, Fan Zhou and Yaowu Chen.

Slides:



Advertisements
Similar presentations
Packet Video Error Concealment With Auto Regressive Model Yongbing Zhang, Xinguang Xiang, Debin Zhao, Siwe Ma, Student Member, IEEE, and Wen Gao, Fellow,
Advertisements

Wen-Hsiao Peng Chun-Chi Chen
Tae-Shick Wang; Kang-Sun Choi; Hyung-Seok Jang; Morales, A.W.; Sung-Jea Ko; IEEE Transactions on Consumer Electronics, Vol. 56, No. 2, May 2010 ENHANCED.
Low-complexity merge candidate decision for fast HEVC encoding Multimedia and Expo Workshops (ICMEW), 2013 IEEE International Conference on Muchen LI,
Adaptive Fast Block-Matching Algorithm by Switching Search Patterns for Sequences With Wide-Range Motion Content 韋弘
Adviser : Ming-Yuan Shieh Student ID : M Student : Chung-Chieh Lien VIDEO OBJECT SEGMENTATION AND ITS SALIENT MOTION DETECTION USING ADAPTIVE BACKGROUND.
A Highly Parallel Framework for HEVC Coding Unit Partitioning Tree Decision on Many-core Processors Chenggang Yan, Yongdong Zhang, Jizheng Xu, Feng Dai,
Efficient Bit Allocation and CTU level Rate Control for HEVC Picture Coding Symposium, 2013, IEEE Junjun Si, Siwei Ma, Wen Gao Insitute of Digital Media,
1 Adaptive slice-level parallelism for H.264/AVC encoding using pre macroblock mode selection Bongsoo Jung, Byeungwoo Jeon Journal of Visual Communication.
Yung-Lin Huang, Yi-Nung Liu, and Shao-Yi Chien Media IC and System Lab Graduate Institute of Networking and Multimedia National Taiwan University Signal.
K.-S. Choi and S.-J. Ko Sch. of Electr. Eng., Korea Univ., Seoul, South Korea IEEE, Electronics Letters Issue Date : June Hierarchical Motion Estimation.
Temporal Video Denoising Based on Multihypothesis Motion Compensation Liwei Guo; Au, O.C.; Mengyao Ma; Zhiqin Liang; Hong Kong Univ. of Sci. & Technol.,
An Improved 3DRS Algorithm for Video De-interlacing Songnan Li, Jianguo Du, Debin Zhao, Qian Huang, Wen Gao in IEEE Proc. Picture Coding Symposium (PCS),
A New Block Based Motion Estimation with True Region Motion Field Jozef Huska & Peter Kulla EUROCON 2007 The International Conference on “Computer as a.
Limin Liu, Member, IEEE Zhen Li, Member, IEEE Edward J. Delp, Fellow, IEEE CSVT 2009.
Light Field Compression Using 2-D Warping and Block Matching Shinjini Kundu Anand Kamat Tarcar EE398A Final Project 1 EE398A - Compression of Light Fields.
Fast Mode Decision for Multiview Video Coding Liquan Shen, Tao Yan, Zhi Liu, Zhaoyang Zhang, Ping An, Lei Yang ICIP
A Fast and Efficient Multi-View Depth Image Coding Method Based on Temporal and Inter- View Correlations of Texture Images Jin Yong Lee Ho Chen Wey Du.
CABAC Based Bit Estimation for Fast H.264 RD Optimization Decision
Video Coding with Spatio-temporal Texture Synthesis and Edge-based inpainting Chunbo Zhu, Xiaoyan Sun, Feng Wu, and Houqiang Li ICME 2008.
2009/04/07 Yun-Yang Ma.  Overview  What is CUDA ◦ Architecture ◦ Programming Model ◦ Memory Model  H.264 Motion Estimation on CUDA ◦ Method ◦ Experimental.
Yen-Lin Lee and Truong Nguyen ECE Dept., UCSD, La Jolla, CA Method and Architecture Design for Motion Compensated Frame Interpolation in High-Definition.
Wei Zhu, Xiang Tian, Fan Zhou and Yaowu Chen IEEE TCE, 2010.
PREDICTIVE 3D SEARCH ALGORITHM FOR MULTI-FRAME MOTION ESTIMATION Lim Hong Yin, Ashraf A. Kassim, Peter H.N de With IEEE Transaction on Consumer Electronics,2008.
Yu-Han Chen, Tung-Chien Chen, Chuan-Yung Tsai, Sung-Fang Tsai, and Liang-Gee Chen, Fellow, IEEE IEEE CSVT
Novel Point-Oriented Inner Searches for Fast Block Motion Lai-Man Po, Chi-Wang Ting, Ka-Man Wong, and Ka-Ho Ng IEEE TRANSACTIONS ON MULTIMEDIA, VOL.9,
Outline Introduction Introduction Fast Inter Prediction Mode Decision for H.264 – –Pre-encoding An Efficient Inter Mode Decision Approach for H.264 Video.
Efficient multi-frame motion estimation algorithms for MPEG-4 AVC/JVTH.264 Mei-Juan Chen, Yi-Yen Chiang, Hung- Ju Li and Ming-Chieh Chi ISCAS 2004.
Efficient Moving Object Segmentation Algorithm Using Background Registration Technique Shao-Yi Chien, Shyh-Yih Ma, and Liang-Gee Chen, Fellow, IEEE Hsin-Hua.
Low-complexity mode decision for MVC Liquan Shen, Zhi Liu, Ping An, Ran Ma and Zhaoyang Zhang CSVT
New Sorting-Based Lossless Motion Estimation Algorithms and a Partial Distortion Elimination Performance Analysis Bartolomeo Montrucchio and Davide Quaglia.
Probabilistic video stabilization using Kalman filtering and mosaicking.
1 Single Reference Frame Multiple Current Macroblocks Scheme for Multiple Reference IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY Tung-Chien.
FAST MULTI-BLOCK SELECTION FOR H.264 VIDEO CODING Chang, A.; Wong, P.H.W.; Yeung, Y.M.; Au, O.C.; Circuits and Systems, ISCAS '04. Proceedings of.
Motion Vector Refinement for High-Performance Transcoding Jeongnam Youn, Ming-Ting Sun, Fellow,IEEE, Chia-Wen Lin IEEE TRANSACTIONS ON MULTIMEDIA, MARCH.
A New Diamond Search Algorithm for Fast Block- Matching Motion Estimation Shan Zhu and Kai-Kuang Ma IEEE TRANSACTIONS ON IMAGE PROCESSION, VOL. 9, NO.
1 An Efficient Mode Decision Algorithm for H.264/AVC Encoding Optimization IEEE TRANSACTION ON MULTIMEDIA Hanli Wang, Student Member, IEEE, Sam Kwong,
IEEE TRANSACTIONS ON IMAGE PROCESSING, VOL. 20, NO. 11, NOVEMBER 2011 Qian Zhang, King Ngi Ngan Department of Electronic Engineering, the Chinese university.
A Low-Power VLSI Architecture for Full-Search Block-Matching Motion Estimation Viet L. Do and Kenneth Y. Yun IEEE Transactions on Circuits and Systems.
1. 1. Problem Statement 2. Overview of H.264/AVC Scalable Extension I. Temporal Scalability II. Spatial Scalability III. Complexity Reduction 3. Previous.
Liquan Shen Zhi Liu Xinpeng Zhang Wenqiang Zhao Zhaoyang Zhang An Effective CU Size Decision Method for HEVC Encoders IEEE TRANSACTIONS ON MULTIMEDIA,
Philipp Merkle, Aljoscha Smolic Karsten Müller, Thomas Wiegand CSVT 2007.
1 Efficient Reference Frame Selector for H.264 Tien-Ying Kuo, Hsin-Ju Lu IEEE CSVT 2008.
Adaptive Multi-path Prediction for Error Resilient H.264 Coding Xiaosong Zhou, C.-C. Jay Kuo University of Southern California Multimedia Signal Processing.
Low-Power H.264 Video Compression Architecture for Mobile Communication Student: Tai-Jung Huang Advisor: Jar-Ferr Yang Teacher: Jenn-Jier Lien.
Fast Mode Decision for H.264/AVC Based on Rate-Distortion Clustering IEEE TRANSACTIONS ON MULTIMEDIA, VOL. 14, NO. 3, JUNE 2012 Yu-Huan Sung Jia-Ching.
2 3 Be introduced in H.264 FRExt profile, but most H.264 profiles do not support it. Do not need motion estimation operation.
Sejong University, DMS Lab. An Efficient True-Motion Estimator Using Candidate Vectors from a Parametric Motion Model Dong-kywn Kim IEEE TRANSACTIONS ON.
Guillaume Laroche, Joel Jung, Beatrice Pesquet-Popescu CSVT
Media Processor Lab. Media Processor Lab. High Performance De-Interlacing Algorithm for Digital Television Displays Media Processor Lab.
Applying 3-D Methods to Video for Compression Salih Burak Gokturk Anne Margot Fernandez Aaron March 13, 2002 EE 392J Project Presentation.
IEEE Transactions on Consumer Electronics, Vol. 58, No. 2, May 2012 Kyungmin Lim, Seongwan Kim, Jaeho Lee, Daehyun Pak and Sangyoun Lee, Member, IEEE 報告者:劉冠宇.
An Effective Three-step Search Algorithm for Motion Estimation
1/39 Motion Adaptive Search for Fast Motion Estimation 授課老師:王立洋老師 製作學生: M 蔡鐘葳.
Outline  Introduction  Observations and analysis  Proposed algorithm  Experimental results 2.
An H.264-based Scheme for 2D to 3D Video Conversion Mahsa T. Pourazad Panos Nasiopoulos Rabab K. Ward IEEE Transactions on Consumer Electronics 2009.
Shen-Chuan Tai, Chien-Shiang Hong, Cheng-An Fu National Cheng Kung University, Tainan City,Taiwan (R.O.C.),DCMC Lab Pacific-Rim Symposium on Image and.
A Fast Video Noise Reduction Method by Using Object-Based Temporal Filtering Thou-Ho (Chao-Ho) Chen, Zhi-Hong Lin, Chin-Hsing Chen and Cheng-Liang Kao.
Motion tracking TEAM D, Project 11: Laura Gui - Timisoara Calin Garboni - Timisoara Peter Horvath - Szeged Peter Kovacs - Debrecen.
Fine-granular Motion Matching for Inter-view Motion Skip Mode in Multi-view Video Coding Haitao Yanh, Yilin Chang, Junyan Huo CSVT.
Fast disparity motion estimation in MVC based on range prediction Xiao Zhong Xu, Yun He ICIP 2008.
Sejong University, DMS Lab. Ki-Hun Han AN EFFECTIVE DE-INTERACING TECHNIQUE USING MOTION COMPENSATED INTERPOLATION IEEE TRANSACTION ON Consumer Electronics,
Presenting: Shlomo Ben-Shoshan, Nir Straze Supervisors: Dr. Ofer Hadar, Dr. Evgeny Kaminsky.
Local Stereo Matching Using Motion Cue and Modified Census in Video Disparity Estimation Zucheul Lee, Ramsin Khoshabeh, Jason Juang and Truong Q. Nguyen.
Complexity varying intra prediction in H.264 Supervisors: Dr. Ofer Hadar, Mr. Evgeny Kaminsky Students: Amit David, Yoav Galon.
Adaptive Block Coding Order for Intra Prediction in HEVC
Fast Decision of Block size, Prediction Mode and Intra Block for H
Fully Scalable Multiview Wavelet Video Coding
Bongsoo Jung, Byeungwoo Jeon
A Block Based MAP Segmentation for Image Compression
Presentation transcript:

{ Fast Disparity Estimation Using Spatio- temporal Correlation of Disparity Field for Multiview Video Coding Wei Zhu, Xiang Tian, Fan Zhou and Yaowu Chen IEEE TCE 2010

 Introduction  Analysis of Disparity Field in MVC  Proposed Fast Disparity Estimation Algorithm  Experimental Results  Conclusion Outline 2

Fig. 1. MVC coding schemes among views for different sequences. Fig. 1. MVC coding schemes among views for different sequences. (a) For “Ballroom” sequence with general 1D camera setup. (b) For “Akko&Kayo” sequence with 2D-array camera setup. (c) For “Flamenco2" sequence with 2D-cross camera setup. Introduction 3

 Motion Estimation  Temporal motion is dependent on the movement of objects.  Only moving objects have motion displacements, objects in background often have no motion displacement.  Disparity Estimation  Inter-view disparity is dependent on the depth of object and the camera setup.  Objects with no motion displacements may also have large disparity displacements. Introduction 4

Analysis of Disparity Field in MVC 5 [17] X. M. Li, D. B. Zhao, S. W. Ma and W. Gao, “Fast disparity and motion estimation based on correlations for multiview video coding,” IEEE Trans. Consumer Electron., vol. 54, no. 4, pp , Nov object depth camera parameters

 Disparity is related to the depth of objects  Shallow depths  large disparity  Distant depths  small disparity  Objects in MVC have large range of depths, the range of disparity is also large.  DE requires a large search range to find optimal disparity vector(DV). Analysis of Disparity Field in MVC 6

7 Fig. 2. Histogram of horizontal disparity Fig. 2. Histogram of horizontal disparity vectors between view S1 and view S0 for vectors between view S1 and view S0 for “Ballroom” sequence. “Ballroom” sequence.  Size of DV vary a lot:  Most of the DV are in [0, 24], background objects.  Some DV are in [24, 72], foreground objects.  Most of DV are in the positive.  The disparity direction is only determined by the location of views.

 Spatial direction correlation Fig. 3. A field of original disparity vectors for the 123th frame of view S1 in Fig. 3. A field of original disparity vectors for the 123th frame of view S1 in “Ballroom” sequence. “Ballroom” sequence. [14] Y. Kim, J. Kim, and K. Sohn, “Fast disparity and motion estimation for multi-view video [14] Y. Kim, J. Kim, and K. Sohn, “Fast disparity and motion estimation for multi-view video coding,” IEEE Trans. Consumer Electron., vol. 53, no.2, pp , May coding,” IEEE Trans. Consumer Electron., vol. 53, no.2, pp , May Analysis of Disparity Field in MVC 8 DVs are highly correlated with neighboring vectors in spatial direction.[14] Some irregular DVs are obvious different with their neighbors, especially in homogenous regions.

 Temporal direction correlation Fig. 4. Histogram of temporal difference for horizontal disparity vectors Fig. 4. Histogram of temporal difference for horizontal disparity vectors between view S1 and view S0 in “Ballroom” sequence. between view S1 and view S0 in “Ballroom” sequence. Analysis of Disparity Field in MVC 9 DVs in the temporal direction are also highly correlated.

 The characteristics of disparity field in MVC:  The size of DVs could vary a lot because of different depths of objects. Due to the fixed camera setup, most of DVs have a consistent direction with the real disparities.  Some DVs deviate from the real disparities, DVs of previous coded should be filtered to be consistent with real disparities.  Since DVs have a high correlation with neighboring vectors in the temporal and spatial direction, the search center of DVs and preliminary DV can be predicted. Analysis of Disparity Field in MVC 10

 Basic idea: Select the search center by using the spatio-temporal correlation of disparity field, and to predict the search range adaptively according to the temporal variation of disparity field.  Part A: Temporal Prediction of the Disparity Vector Based on the Smoothed Disparity Field of the Previous Coded Frame.  Part B: Selection of Search Center Based on the Spatial-temporal Correlation of Disparity Field.  Part C: Prediction of Search Range Based on the Temporal Variation of Disparity Field.  Part D: Overall Proposed Algorithm. Proposed Fast Disparity Estimation Algorithm 11

 Basic idea: There is a high correlation of DVs in temporal direction. Use previous coded frames to predict current frame.  Problem: Some noisy DVs are not consistent with their real disparities, these vectors should be eliminated to obtain a smooth disparity field.  Solution: Because noisy DVs have irregular directions, check every DV direction first.  Requirement: GDV is calculated by averaging all DVs of block 16x16 in the previous coded frame, and its direction is used as the reference of the real disparity direction. Part A: Temporal Prediction of the Disparity Vector Based on the Smoothed Disparity Field of the Previous Coded Frame 12

 Step1: Using GDV, the DV of block 16x16 for each MB is regularized. Part A: Temporal Prediction of the Disparity Vector Based on the Smoothed Disparity Field of the Previous Coded Frame 13

Part A: Temporal Prediction of the Disparity Vector Based on the Smoothed Disparity Field of the Previous Coded Frame 14

Part A: Temporal Prediction of the Disparity Vector Based on the Smoothed Disparity Field of the Previous Coded Frame 15 encoding order of current frame 0.5 co-located (Zero Disparity Vector)

 Basic idea: The disparity vectors are highly correlated in spatio- temporal directions, so the neighboring disparity vectors in spatio- temporal directions are used to determine the search center for the current block.  Select candidates:  The neighboring disparity vectors are selected as candidates of the search center.  Non-anchor frame: 5-9 are selected from the forward and backward temporal reference frame.  Anchor frame: 5-9 are selected from the previous coded anchor frame. Part B: Selection of Search Center Based on the Spatial-temporal Correlation of Disparity Field 16

Part B: Selection of Search Center Based on the Spatial-temporal Correlation of Disparity Field 17

Part B: Selection of Search Center Based on the Spatial-temporal Correlation of Disparity Field 18 It’s the smallest, only a smaller search range is needed.

 Temporal variation(distance) of DV: |CDV – TDV|  CDV: An approximation of the DV in current frame.  TDV: A temporal prediction of the DV in previous coded frame.  Basic idea: the distance is related the DV consistency between current frame and previous coded frame.  Small distance  small search range  Large distance  large search range Part C: Prediction of Search Range Based on the Temporal Variation of Disparity Field. 19

 Search range: Part C: Prediction of Search Range Based on the Temporal Variation of Disparity Field. 20 (larger)  Calculate search range

Part D: Overall Proposed Algorithm 21

 JMVC4.0  Testing Configuration  Only Inter16x16 mode for inter mode  Use first two or three views of the sequences.  View S1 was chosen for proposed algorithm, and disable ME  S0 was chosen as the reference view for Flamenco2(2D-cross), S0 and S2 were chosen for others.  Comparing with full search algorithm and fast search algorithm. Experimental Results 22

 Rate-distortion performance Experimental Results 23

 Performance comparison Experimental Results 24

 In this paper, a fast DE algorithm is proposed to save the computational load of MVC.  Take into account the spatial-temporal correlation and the temporal variation of disparity.  Perform well on all test sequences.  Compare with full search algorithm, achieve an average 96% reduction of computational complexity, while RD performance remain the same.  Compare with fast search algorithm, achieve an average 43% reduction of computational complexity, while RD performance is improved. Conclusion 25