光照變化下基于逆向稀疏表示的視覺跟蹤方法
doi: 10.11999/JEIT180442
-
大連大學(xué)信息工程學(xué)院 ??大連 ??116622
Visual Tracking Method Based on Reverse Sparse Representation under Illumination Variation
-
College of Information Engineering, Dalian University, Dalian 116622, China
-
摘要:
針對光照變化引起目標(biāo)跟蹤性能顯著下降的問題，該文提出一種聯(lián)合優(yōu)化光照補(bǔ)償和多任務(wù)逆向稀疏表示的視覺跟蹤方法。首先基于模板與候選目標(biāo)的平均亮度差異對模板實(shí)施光照補(bǔ)償，并利用候選目標(biāo)逆向稀疏表示光照補(bǔ)償后的模板。而后將所得多個關(guān)于單模板的優(yōu)化問題轉(zhuǎn)化為一個關(guān)于多模板的多任務(wù)優(yōu)化問題，并利用交替迭代方法求解此多任務(wù)優(yōu)化問題以獲得最優(yōu)光照補(bǔ)償系數(shù)矩陣以及稀疏編碼矩陣。最后利用所得稀疏編碼矩陣快速剔除無關(guān)候選目標(biāo)，并采用局部結(jié)構(gòu)化評估方法實(shí)現(xiàn)目標(biāo)精確跟蹤。仿真結(jié)果表明，與現(xiàn)有主流算法相比，劇烈光照變化情況下，所提方法可顯著改善目標(biāo)跟蹤精度及穩(wěn)健性。
Abstract:Focusing on the issue of heavy decrease of object tracking performance induced by illumination variation, a visual tracking method via jointly optimizing the illumination compensation and multi-task reverse sparse representation is proposed. The template illumination is firstly compensated by the developed algorithm, which is based on the average brightness difference between templates and candidates. In what follows, the candidate set is exploited to sparsely represent the templates after illumination compensation. Subsequently, the obtained multiple optimization issues associated with single template can be recast as a multi-task optimization one related to multiple templates, which can be solved by the alternative iteration approach to acquire the optimal illumination compensation coefficient and the sparse coding matrix. Finally, the obtained sparse coding matrix can be exploited to quickly eliminate the unrelated candidates, afterwards the local structured evaluation method is employed to achieve the accurate object tracking. As compared to the existing state-of-the-art algorithms, simulation results show that the proposed algorithm can improve the accuracy and robustness of the object tracking significantly in the presence of heavy illumination variation.
-
Key words:
- Visual tracking /
- Illumination compensation /
- Sparse representation /
- Particle filter
-
表 1 光照補(bǔ)償與多任務(wù)逆向稀疏表示聯(lián)合優(yōu)化算法
輸入：${T}$, ${Y}$, $\beta $和$\tilde \lambda $ (1) 基于式(8)設(shè)定稀疏編碼矩陣${C}$的初始值； (2) 由式(12)，式(2)，式(4)，式(6)獲得${K}$； (3) 利用APG方法求解問題式(13)以求得${C}$； (4) 重復(fù)步驟(2)，步驟(3)，直至滿足收斂條件。 輸出：${K}$和${C}$ 
下載: 導(dǎo)出CSV
表 2 視頻序列及其主要挑戰(zhàn)
測試序列 挑戰(zhàn)因素 Car4 光照變化，尺度變化 Singer1 光照變化，尺度變化，遮擋等 Trellis 光照變化，背景雜波，尺度變化等 Car1 光照變化，運(yùn)動模糊，尺度變化等
下載: 導(dǎo)出CSV
表 3 不同跟蹤方法的平均中心位置誤差和平均跟蹤重疊率
測試序列 平均中心位置誤差(像素) 平均跟蹤重疊率 本文 TLD Struck L1APG MTT 本文 TLD Struck L1APG MTT Car4 3.47 12.84 8.69 77.00 22.34 0.84 0.63 0.49 0.25 0.45 Singer1 2.88 7.99 14.51 53.35 36.17 0.86 0.73 0.36 0.28 0.34 Trellis 6.82 31.06 6.92 62.20 68.80 0.65 0.48 0.61 0.20 0.21 Car1 1.18 85.15 51.73 93.93 101.81 0.83 0.26 0.11 0.17 0.15 平均 3.59 24.26 20.46 71.62 57.28 0.80 0.53 0.40 0.23 0.29
下載: 導(dǎo)出CSV
表 4 快速候選目標(biāo)篩選方案對運(yùn)行速度(FPS)的影響
測試序列 Car4 Singer1 Trellis Car1 不采用篩選方案運(yùn)行速度(FPS) 4.1 4.6 3.1 5.5 采用篩選方案運(yùn)行速度(FPS) 10.5 8.7 10.4 8.4
下載: 導(dǎo)出CSV
-
FRADI H, LUVISON B, and PHAM Q C. Crowd behavior analysis using local mid-level visual descriptors[J]. IEEE Transactions on Circuits & Systems for Video Technology, 2017, 27(3): 589–602. doi: 10.1109/TCSVT.2016.2615443 YU Gang, LI Chao, and SHANG Zeyuan. Video monitoring method, video monitoring system and computer program product[P]. USA Patent, 9792505, 2017. UENG S K and CHEN Guanzhi. Vision based multi-user human computer interaction[J]. Multimedia Tools & Applications, 2016, 75(16): 10059–10076. doi: 10.1007/s11042-015-3061-z WU Yi, LIM J, and YANG Minghsuan. Object tracking benchmark[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2015, 37(9): 1834–1848. doi: 10.1109/TPAMI.2014.2388226 PAN Zheng, LIU Shuai, and FU Weina. A review of visual moving target tracking[J]. Multimedia Tools & Applications, 2017, 76(16): 16989–17018. doi: 10.1007/s11042-016-3647-0 薛模根, 劉文琢, 袁廣林, 等. 基于編碼遷移的快速魯棒視覺跟蹤[J]. 電子與信息學(xué)報, 2017, 39(7): 1571–1577. doi: 10.11999/JEIT160966XUE Mogen, LIU Wenzhuo, YUAN Guanglin, et al. Fast robust visual tracking based on coding transfer[J]. Journal of Electronics &Information Technology, 2017, 39(7): 1571–1577. doi: 10.11999/JEIT160966 楊峰, 張婉瑩. 一種多模型貝努利粒子濾波機(jī)動目標(biāo)跟蹤算法[J]. 電子與信息學(xué)報, 2017, 39(3): 634–639. doi: 10.11999/JEIT160467YANG Feng and ZHANG Wanying. Multiple model Bernoulli particle filter for maneuvering target tracking[J]. Journal of Electronics &Information Technology, 2017, 39(3): 634–639. doi: 10.11999/JEIT160467 BAIG M Z and GOKHALE A V. Object tracking using mean shift algorithm with illumination invariance[C]. Fifth International Conference on Communication Systems and Network Technologies, Gwalior, India, 2015: 550–553. NAYAK A and CHAUDHURI S. Automatic illumination correction for scene enhancement and object tracking[J]. Image & Vision Computing, 2006, 24(9): 949–959. doi: 10.1016/j.imavis.2006.02.017 SILVEIRA G and MALIS E. Real-time visual tracking under arbitrary illumination changes[C]. IEEE Conference on Computer Vision and Pattern Recognition, Minneapolis, USA, 2007: 1–6. WANG Yuru, TANG Xianglong, CUI Qing, et al. Dynamic appearance model for particle filter based visual tracking[J]. Pattern Recognition, 2012, 45(12): 4510–4523. doi: 10.1016/j.patcog.2012.05.010 BAO Chenglong, WU Yi, LING Haibin, et al. Real time robust L1 tracker using accelerated proximal gradient approach[C]. IEEE Conference on Computer Vision and Pattern Recognition, Providence, USA, 2012: 1830–1837. MA Bo, SHEN Jianbing, LIU Yangbiao, et al. Visual tracking using strong classifier and structural local sparse descriptors[J]. IEEE Transactions on Multimedia, 2015, 17(10): 1818–1828. doi: 10.1109/TMM.2015.2463221 ZHUANG Bohan, LU Huchuan, XIAO Ziyang, et al. Visual tracking via discriminative sparse similarity map[J]. IEEE Transactions on Image Processing, 2014, 23(4): 1872–1881. doi: 10.1109/TIP.2014.2308414 JIA Xu, LU Huchuan, and YANG Minghsuan. Visual tracking via coarse and fine structural local sparse appearance models[J]. IEEE Transactions on Image Processing, 2016, 25(10): 4555–4564. doi: 10.1109/TIP.2016.2592701 SUI Yao and ZHANG Li. Robust tracking via locally structured representation[J]. International Journal of Computer Vision, 2016, 119(2): 110–144. doi: 10.1007/s11263-016-0881-x ZHANG Tianzhu, GHANEM B, LIU Si, et al. Robust visual tracking via multi-task sparse learning[C]. IEEE Conference on Computer Vision and Pattern Recognition, Providence, USA, 2012: 2042–2049. MA Bo, HUANG Lianghua, SHEN Jianbing, et al. Visual tracking under motion blur[J]. IEEE Transactions on Image Processing, 2016, 25(12): 5867–5876. doi: 10.1109/TIP.2016.2615812 ROSS D A, LIM J, LIN R S, et al. Incremental learning for robust visual tracking[J]. International Journal of Computer Vision, 2008, 77(1): 125–141. doi: 10.1007/s11263-007-0075-7 POLSON N and SOKOLOV V. Bayesian particle tracking of traffic flows[J]. IEEE Transactions on Intelligent Transportation Systems, 2018, 19(2): 345–356. doi: 10.1109/TITS.2017.2650947 HE Zhenyu, YI Shuangyan, CHEUNG Y M, et al. Robust object tracking via key patch sparse representation[J]. IEEE Transactions on Cybernetics, 2017, 47(2): 354–364. doi: 10.1109/TCYB.2016.2514714 ZHANG Kaihua, ZHANG Lei, and YANG Minghsuan. Real-time compressive tracking[C]. European Conference on Computer Vision, Florence, Italy, 2012: 864–877. KALAL Z, MATAS J, and MIKOLAJCZYK K. P-N learning: Bootstrapping binary classifiers by structural constraints[C]. IEEE Conference on Computer Vision and Pattern Recognition, San Francisco, USA, 2010: 49–56. HARE S, SAFFARI A, and TORR P H S. Struck: Structured output tracking with kernels[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2016, 38(10): 2096–2109. doi: 10.1109/TPAMI.2015.2509974 -
計量
- 文章訪問數(shù): 1826
- HTML全文瀏覽量: 602
- PDF下載量: 78
- 被引次數(shù): 0
下載:
