顯著中心先驗和顯著-深度概率矯正的RGB-D顯著目標檢測

劉政怡; 黃子超; 張志華

doi:10.11999/JEIT170235

顯著中心先驗和顯著-深度概率矯正的RGB-D顯著目標檢測

doi: 10.11999/JEIT170235

基金項目:

國家科技支撐計劃(2015BAK24B00)，安徽高校省級自然科學研究項目(KJ2015A009)，安徽大學信息保障技術協(xié)同創(chuàng)新中心開放課題

計量
- 文章訪問數(shù): 1379
- HTML全文瀏覽量: 249
- PDF下載量: 260
- 被引次數(shù): 0
出版歷程
- 收稿日期: 2017-03-20
- 修回日期: 2017-07-04
- 刊出日期: 2017-12-19

RGB-D Saliency detection Based on Saliency Center Prior and Saliency-depth Probability Adjustment

Funds:

The National Key Technology RD Program of the Ministry of Science and Technology of China (2015BAK24B00), The Key Program of Natural Science Project of Educational Commission of Anhui Province (KJ2015A009), The Open Issues on Co-Innovation Center for Information Supply Assurance Technology, Anhui University

摘要

摘要: 隨著深度特征在圖像顯著檢測領域中發(fā)揮越來越重要的作用，傳統(tǒng)的RGB圖像顯著檢測模型由于未能充分利用深度信息已經(jīng)不能適用于RGB-D圖像的顯著檢測。該文提出顯著中心先驗和顯著-深度(S-D)概率矯正的RGB-D顯著檢測模型，使得深度特征和RGB特征間相互指導，相互補充。首先，依據(jù)3維空間權重和深度先驗獲取深度圖像初步顯著圖；其次，采用特征融合的流形排序算法獲取RGB圖像的初步顯著圖。接著，計算基于深度的顯著中心先驗，并以該先驗作為顯著權重進一步提升RGB圖像的顯著檢測結果，獲取RGB圖像最終顯著圖；再次，計算顯著-深度矯正概率，并對深度圖的初步顯著檢測結果使用此概率進行矯正。接著，計算基于RGB的顯著中心先驗，并以該先驗作為顯著權重進一步提升深度圖像矯正后的顯著檢測結果，獲取深度圖像的最終顯著圖；最后，采用優(yōu)化框架對深度圖像最終顯著圖進行優(yōu)化得到RGB-D圖像最終的顯著圖。所有的對比實驗都是在公開的數(shù)據(jù)集NLPR RGBD-1000數(shù)據(jù)集上進行，實驗結果顯示該文算法較當前流行的算法有更好的性能。
- 3維空間權重 /
- 特征融合的流形排序算法 /
- 顯著中心先驗 /
- 顯著-深度概率矯正
Abstract: Along with more and more important role of depth features played in computer saliency community, traditional RGB saliency models can not directly utilized for saliency detection on RGB-D domains. This paper proposes saliency center prior and Saliency-Depth (S-D) probability adjustment RGB-D saliency detection framework, making the depth and RGB features adaptively fuse and complementary to each other. First, the initial saliency maps of depth images are obtained according to three-dimension space weights and depth prior; second, the feature fused Manifold Ranking model with extracted depth features is utilized for RGB image saliency detection. Then, the saliency center prior based on depth is computed and this value is used as saliency weight to further improve the RGB image saliency detection results, obtaining the final RGB saliency map. After that, Saliency-Depth (S-D) rectify probability is also computed and the saliency results of depth images are corrected with this probability. Then the saliency center prior based on RGB is also computed and this value is used as saliency weights to further improve the depth image saliency detection results and to obtain the final depth saliency maps. Finally the optimization framework is utilized to optimize the depth image final saliency maps and to obtain the final RGB-D saliency map. All the experiments are executed on the public NLPR RGBD-1000 benchmark and extensive experiments demonstrate that the proposed algorithm achieves better performance compared with existing state-of-the-art approaches.
- Three-dimension space weights /
- Feature fused driven manifold ranking /
- Saliency center prior /
- Saliency- Depth (S-D) probability adjustment

HTML全文

參考文獻(22)

DING Y Y, XIAO J, and YU J Y. Importance filtering for image retargeting[C]. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Providence, RI, USA, 2011: 89-96. doi: 10.1109/CVPR.2011.5995445.

DONOSER M, URSCHLER M, HIRZER M, et al. Saliency driven total variation segmentation[C]. IEEE International Conference on Computer Vision, Kyoto, 2009: 817-824. doi: 10.1109/ICCV.2009.5459296.

SRIVASTAVA S, MUKHERJEE P, and LALL B. Adaptive image compression using saliency and KAZE features[C]. International Conference on Signal Processing and Communications, Bangalore, 2016: 1-5. doi: 10.1109/ SPCOM.2016.7746680.

SIAGIAN C and ITTI L. Rapid biologically-inspired scene classification using features shared with visual attention[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2007, 29(2): 300-312. doi: 10.1109/TPAMI.2007. 40.

WANG X J, MA W Y, and LI X. Data-driven approach for bridging the cognitive gap in image retrieval[C]. IEEE International Conference on Multimedia and Expo, Taipei, 2004, 3: 2231-2234. doi: 10.1109/ICME.2004.1394714.

MAHADEVAN V and VASCONCELOS N. Saliency-based discriminant tracking[C]. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Miami, Florida, USA, 2009: 1007-1013. doi: 10.1109/CVPR. 2009.5206573.

REN J, GONG X, YU L, et al. Exploiting global priors for RGB-D saliency detection[C]. IEEE Conference on Computer Vision and Pattern Recognition Workshops, Boston, MA, 2015: 25-32. doi: 10.1109/CVPRW.2015.7301391.

LI W, QIU J, and LI X. Visual saliency detection based on gradient contrast and color complexity[C]. International Conference on Internet Multimedia Computing and Service, Zhangjiajie, China, 2015: 1-5. doi: 10.1145/2808492.2808534.

ZHU H, SHENG B, LIN X, et al. Foreground object sensing for saliency detection[C]. ACM on International Conference on Multimedia Retrieval, New York, USA, 2016: 111-118. doi: 10.1145/2911996.2912008.

WANG T, ZHANG L, LU H, et al. Kernelized subspace ranking for saliency detection[C]. European Conference on Computer Vision, Springer International Publishing, 2016: 450-466. doi: 10.1007/978-3-319-46484-8_27.

QIN Y, LU H, XU Y, et al. Saliency detection via cellular automata[C]. IEEE Conference on Computer Vision and Pattern Recognition, Boston, MA, 2015: 110-119. doi: 10.1109/CVPR.2015.7298606.

LANG C, NGUYEN T V, KATTI H, et al. Depth matters: Influence of depth cues on visual saliency[J]. Lecture Notes in Computer Science, 2012(2): 101-115. doi: 10.1007/978-3-642- 33709-3_8.

YANG C, ZHANG L, LU H, et al. Saliency detection via graph-based Manifold Ranking[C]. IEEE Computer Vision and Pattern Recognition, Portland, OR, 2013: 3166-3173. doi: 10.1109/CVPR.2013.407.

GUO J, REN T, BEI J, et al. Salient object detection in RGB-D image based on saliency fusion and propagation[C]. ACM International Conference on Internet Multimedia Computing and Service, Zhangjiajie, China, 2015: 59-63. doi: 10.1145/2808492.2808551.

DESINGH K, MADHAVA K K, RAJAN D, et al. Depth really matters: improving visual salient region detection with depth[C]. British Machine Vision Conference, Bristol, 2013: 98.1-98.11. doi: 10.5244/C.27.98.

CHENG M M, MITRA N J, HUANG X, et al. Global contrast based salient region detection[J]. IEEE Transactions on Pattern Analysis Machine Intelligence, 2015, 37(3): 569-582. doi: 10.1109/TPAMI.2014.2345401.

YANG C, ZHANG L, and LU H. Graph-regularized saliency detection with convex-hull-based center prior[J]. IEEE Signal Processing Letters, 2013, 20(7): 637-640. doi: 10.1109/LSP. 2013.2260737.

HAREL J, KOCH C, and Perona P. Graph-based visual saliency[C]. Proceedings of the Twentieth Annual Conference on Neural Information Processing Systems, Vancouver, British Columbia, Canada, 2006: 545-552.

ZHU W, LIANG S, Wei Y, et al. Saliency Optimization from robust background detection[C]. IEEE Conference on Computer Vision and Pattern Recognition, Columbus, OH, 2014: 2814-2821. doi: 10.1109/CVPR.2014.360.

PENG H, LI B, XIONG W, et al. RGBD salient object detection: A benchmark and algorithms[J]. Lecture Notes in Computer Science, 2014, 8691: 92-109. doi: 10.1007/978-3- 319-10578-9_7.

JU R, LIU Y, REN T, et al. Depth-aware salient object detection using anisotropic center-surround difference[J]. Image Communication, 2015, 38(C): 115-126. doi: 10.1016/ j.image.2015.07.002.

CHENG Y, FU H, WEI X, et al. Depth enhanced saliency detection method[C]. Proceedings of International Conference on Internet Multimedia Computing and Service, Xiamen, 2014: 23-27. doi: 10.1145/2632856.2632866.

施引文獻

資源附件(0)

訪問統(tǒng)計