Hyperspectral Target Tracking Based on Dimensionality Reduction of Structural Tensors and Improved Context-Aware Correlation Filter

Dong Zhao; Bin Hu; Yuchen Zhuang; Xiang Teng; Chao Wang; Jia Li; Yecai Guo

doi:10.3788/AOS240464

Acta Optica Sinica, Volume. 44, Issue 11, 1115001(2024)

Hyperspectral Target Tracking Based on Dimensionality Reduction of Structural Tensors and Improved Context-Aware Correlation Filter

Dong Zhao^1,2, Bin Hu^1,2, Yuchen Zhuang¹, Xiang Teng³, Chao Wang¹, Jia Li⁴, and Yecai Guo^1,2、*

Author Affiliations

¹School of Electronics & Information Engineering, Nanjing University of Information Science & Technology, Nanjing 210044, Jiangsu, China

²School of Electronics and Information Engineering, Wuxi University, Wuxi 214105, Jiangsu, China

³School of Physics, Xidian University, Xian710071, Shaanxi, China

⁴Department of Basic Sciences, Air Force Engineering University, Xi an 710051, Shaanxi, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(13)

Fig. 1. General framework of SI-HVT algorithm

Download full size

Fig. 2. Experimental results of dimensionality reduction algorithms. (a) Original hyperspectral image; (b) target structure tensor (ST) image; (c) dimension reduction result image

Download full size

Fig. 3. Experimental results of adaptive tracking response map. (a) Initial frame response map; (b) response map with target occluded; (c) response map after fusion

Download full size

Fig. 4. Qualitative comparison results of four video sequences. (a) Ball; (b) toy 1; (c) bus 2; (d) kangroo

Download full size

Fig. 5. Evaluation curves of different tracking algorithms on test set. (a) Accuracy chart; (b) success rate chart

Download full size

Fig. 6. Evaluation curves of different tracking algorithms under target occlusion challenges. (a) Accuracy chart; (b) success rate chart

Download full size

Fig. 7. Evaluation curves of different tracking algorithms under fast moving challenges. (a) Accuracy chart; (b) success rate chart

Download full size

Fig. 8. Influence of different $N_{O}$ / $H_{O}$ on dimensionality reduction results. (a) Structure similarity chart; (b) peak signal-to-noise ratio chart

Download full size

Fig. 9. Evaluation curves of compared algorithms on test set. (a) Accuracy chart; (b) success rate chart

Download full size

Table 1. Success rate of each algorithm under different challenges
View table
Table 1. Success rate of each algorithm under different challenges
Algorithm Suc_All Suc_OCC Suc_FM
SI-HVT 0.680 0.620 0.754
SiamBAG 0.635 0.616 0.683
MFI 0.604 0.555 0.674
MHT 0.584 0.559 0.626
DEEPHKCF 0.386 0.327 0.494

Table 2. Accuracy of each algorithm under different challenges
View table
Table 2. Accuracy of each algorithm under different challenges
Algorithm Acc_ALL Acc_OCC Acc_FM
SI-HVT 0.899 0.870 0.961
SiamBAG 0.858 0.861 0.947
MFI 0.848 0.861 0.945
MHT 0.845 0.834 0.932
DEEPHKCF 0.676 0.619 0.879

Table 3. Experimental results of compared dimensionality reduction algorithm
View table
Table 3. Experimental results of compared dimensionality reduction algorithm
Performance index Dimensionality reduction algorithm
Proposed PCA KPCA 1*1 convolution kernel
SSIM 0.754 0.540 0.647 0.238
PSNR /dB 7.887 5.385 6.891 5.213

Table 4. Ablation experimenal resultst of each module of SI-HVT
View table
Table 4. Ablation experimenal resultst of each module of SI-HVT
Evaluation indicator Proposed Im-Sample AR PCA
Accuracy 0.899 0.874 0.841 0.661
Success rate 0.680 0.671 0.589 0.396

Tools

Get Citation

Copy Citation Text

Dong Zhao, Bin Hu, Yuchen Zhuang, Xiang Teng, Chao Wang, Jia Li, Yecai Guo. Hyperspectral Target Tracking Based on Dimensionality Reduction of Structural Tensors and Improved Context-Aware Correlation Filter[J]. Acta Optica Sinica, 2024, 44(11): 1115001

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites