Hyperspectral Target Tracking Based on Spectral Matching Dimensionality Reduction and Feature Fusion

Yecai Guo; Jialu Cao; Yingying Han; Tianmeng Zhang; Dong Zhao; Xu Tao

doi:10.3788/AOS230776

Acta Optica Sinica, Volume. 43, Issue 20, 2012002(2023)

Hyperspectral Target Tracking Based on Spectral Matching Dimensionality Reduction and Feature Fusion

Yecai Guo^1,2, Jialu Cao^1,2, Yingying Han³, Tianmeng Zhang⁴, Dong Zhao^1,2、*, and Xu Tao^1,2

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

³No. 703 Research Institute of China State Shipbuilding Corporation Limited, Harbin 150000, Heilongjiang , China

⁴College of Aerospace and Civil Engineering, Harbin Engineering University, Harbin 150000, Heilongjiang , China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(14)

Fig. 1. Overall flow chart of DF-HVT

Download full size

Fig. 2. Dimensionality reduction results. (a) Original hyperspectral image; (b) local spectral curve and average spectral curve; (c) image after dimension reduction

Download full size

Fig. 3. Depth features of the first 32 channels of the bottleneck_13 layer

Download full size

Fig. 4. Schematic of local target enhancement

Download full size

Fig. 5. 3D HOG features of the first 32 channels after enhancing the target

Download full size

Fig. 6. DC-HOG features of the first 32 channels

Download full size

Fig. 7. Qualitative analysis results on the selected sequences. (a) Book sequence; (b) excavator sequence; (c) car sequence; (d) face sequence

Download full size

Fig. 8. Tracking precision and success rate of four algorithms on the test sequences. (a) Precision; (b) success rate

Download full size

Fig. 9. Tracking precision and success rate of four algorithms on the scale variation challenge. (a) Precision; (b) success rate

Download full size

Fig. 10. Tracking precision and success rate of four algorithms on out-of-plane challenge. (a) Precision; (b) success rate

Download full size

Fig. 11. Results of ablation experiments for all test sequences. (a) Precision; (b) success rate

Download full size

Table 1. Precision of four algorithms under different challenges
View table
Table 1. Precision of four algorithms under different challenges
Algorithm Precision Precision with SV Precision with OPR
Ours 0.820 0.905 0.916
MHT 0.787 0.707 0.927
DeepHKCF 0.786 0.693 0.779
CNHT 0.647 0.627 0.418

Table 2. Success rate of four algorithms under different challenges
View table
Table 2. Success rate of four algorithms under different challenges
Algorithm Success rate Success rate with SV Successrate with OPR
Ours 0.641 0.671 0.763
MHT 0.619 0.531 0.736
DeepHKCF 0.617 0.483 0.536
CNHT 0.356 0.440 0.386

Table 3. Precision and success rate of ablation experiment
View table
Table 3. Precision and success rate of ablation experiment
Methods Ours 3D HOG average PCA
Precision 0.820 0.776 0.720 0.676
Success rate 0.641 0.589 0.555 0.532

Tools

Get Citation

Copy Citation Text

Yecai Guo, Jialu Cao, Yingying Han, Tianmeng Zhang, Dong Zhao, Xu Tao. Hyperspectral Target Tracking Based on Spectral Matching Dimensionality Reduction and Feature Fusion[J]. Acta Optica Sinica, 2023, 43(20): 2012002

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites