Consistent Depth Maps Estimation from Binocular Stereo Video Sequence

doi:10.1007/s12204-016-1710-7

Abstract

Abstract: In the paper, an approach is proposed for the problem of consistency in depth maps estimation from binocular stereo video sequence. The consistent method includes temporal consistency and spatial consistency to eliminate the flickering artifacts and smooth inaccuracy in depth recovery. So the improved global stereo matching based on graph cut and energy optimization is implemented. In temporal domain, the penalty function with coherence factor is introduced for temporal consistency, and the factor is determined by Lucas-Kanade optical flow weighted histogram similarity constraint (LKWHSC). In spatial domain, the joint bilateral truncated absolute difference (JBTAD) is proposed for segmentation smoothing. The method can smooth naturally and uniformly in low-gradient region and avoid over-smoothing as well as keep edge sharpness in high-gradient discontinuities to realize spatial consistency. The experimental results show that the algorithm can obtain better spatial and temporal consistent depth maps compared with the existing algorithms.

Key words: consistent depth maps| binocular stereo video sequence| Lucas-Kanade optical flow weighted histogram similarity constraint (LKWHSC)| joint bilateral truncated absolute difference (JBTAD)

摘要： In the paper, an approach is proposed for the problem of consistency in depth maps estimation from binocular stereo video sequence. The consistent method includes temporal consistency and spatial consistency to eliminate the flickering artifacts and smooth inaccuracy in depth recovery. So the improved global stereo matching based on graph cut and energy optimization is implemented. In temporal domain, the penalty function with coherence factor is introduced for temporal consistency, and the factor is determined by Lucas-Kanade optical flow weighted histogram similarity constraint (LKWHSC). In spatial domain, the joint bilateral truncated absolute difference (JBTAD) is proposed for segmentation smoothing. The method can smooth naturally and uniformly in low-gradient region and avoid over-smoothing as well as keep edge sharpness in high-gradient discontinuities to realize spatial consistency. The experimental results show that the algorithm can obtain better spatial and temporal consistent depth maps compared with the existing algorithms.

关键词: consistent depth maps| binocular stereo video sequence| Lucas-Kanade optical flow weighted histogram similarity constraint (LKWHSC)| joint bilateral truncated absolute difference (JBTAD)

CLC Number:

TN 911.73

DUAN Fengfeng (段峰峰). Consistent Depth Maps Estimation from Binocular Stereo Video Sequence[J]. Journal of shanghai Jiaotong University (Science), 2016, 21(2): 184-191.

References 18

[1]	KHOSHABEH R, CHAN S H, NGUYEN T Q.Spatio-temporal consistency in video disparity estimation[C]//Proceedings of IEEE International Conferenceon Acoustics, Speech and Signal Processing.Prague, Czech Republic: IEEE, 2011: 885-888.
[2]	CIGLA C, ALATAN A A. Temporally consistentdense depth map estimation via belief propagation[C]//IEEE 3DTV Conference on the True Vision—Capture, Transmission and Display of 3D Video. Potsdam,Germany: IEEE, 2009: 1-4.
[3]	GARCIA F, AOUADA D, MIRBACH B, et al.A new multi-lateral filter for real-time depth enhancement[C]//IEEE International Conference onAdvanced Video and Signal-Based Surveillance. Klagenfurt,Austria: IEEE, 2011: 42-47.
[4]	RICHARDT C, ORR D, DAVIES I, et al. Realtimespatiotemporal stereo matching using the dualcross-bilateral grid [C]//Proceedings of the EuropeanConference on Computer Vision. Hersonissos, Greece:Springer-Verlag, 2010, 6313: 510-523.
[5]	LEE S B, HO Y S. Temporally consistent depth map estimation for 3D video generation and coding [J].China Communications, 2013, 10(5): 39-49.
[6]	PHAM C C, NGUYEN V D, JEON J W. Efficient spatio-temporal local stereo matching using information permeability filtering [C]//IEEE InternationalConference on Image Processing. Orlando, USA:IEEE, 2012: 2965-2968.
[7]	MIN D B, LU J B, DO M N. Depth video enhancement based on weighted mode filtering [J]. IEEE Transactions on Image Processing, 2012, 21(3): 1176-1190.
[8]	FUSIELLO A, IRSARA L. Quasi-Euclidean uncalibratedepipolar rectification [C]//19th InternationalConference on Pattern Recognition. Tampa, USA:IEEE, 2008: 1-4.
[9]	HEO Y S, LEE K M, LEE S U. Joint depth mapand color consistency estimation for stereo images withdifferent illuminations and cameras [J]. IEEE Transactionson Pattern Analysis and Machine Intelligence,2013, 35(5): 1094-1106.
[10]	BOYKOV Y, VEKSLER O, ZABIH R. Fast approximateenergy minimization via graph cuts [J]. IEEETransactions on Pattern Analysis and Machine Intelligence,2001, 23(11): 1222-1239.
[11]	QIN S, XIE G. LW-PGD method and fusion featurebased real-time pedestrian detection in video [J].Journal of Computational Information Systems, 2014,10(6): 2273-2281.
[12]	JIANG L C, SHEN G Q, ZHANG G X. An imageretrieval algorithm based on HSV color segment histograms[J]. Mechanical & Electrical Engineering Magazine,2009, 26(11): 54-57 (in Chinese).
[13]	ZHANG Y, ZHANG J W, YANG G Q, et al. Videode-hazing using spatial-temporal coherence optimization[J]. Application Research of Computers, 2011,28(10): 3983-3985 (in Chinese).
[14]	BUADES A, COLL B, MOREL J M. Nonlocal imageand movie denoising [J]. International Journal ofComputer Vision, 2008, 76(2): 123-139.
[15]	FERREIRA L, ASSUNCAO P, DA SILVA CRUZ LA. 3D video shot boundary detection based on clusteringof depth-temporal features [C]//2013 11th InternationalWorkshop on Content-based Multimedia Indexing.Veszprem, Hungary: IEEE, 2013: 1-6.
[16]	MA G H, WANG C, LIU P, et al. Sequential similaritydetection algorithm based on image edge feature[J]. Journal of Shanghai Jiaotong University (Science),2014, 19(1): 79-83.
[17]	ZHANG G F, JIA J Y, WONG T T, et al. Consistentdepth maps recovery from a video sequence [J].IEEE Transactions on Pattern Analysis and MachineIntelligence, 2009, 31(6): 974-988.
[18]	YANG Q X, YANG R G, DAVIS J, et al. Spatialdepthsuper resolution for range images [C]//IEEEConference on Computer Vision and Pattern Recognition.Minneapolis, USA: IEEE, 2007: 1-8.