Infrared and Laser Engineering, Volume. 53, Issue 12, 20240324(2024)
Research on biomimetic polarization vision solid/gas heat source identification technology based on deep learning
Figures & Tables(11)
Fig. 2. (a) Orientation of microvilli in the 5th and 6th rows of microphthalmic cells in the left eye of Stomatopods; (b) A three-channel polarization distance model with photoreceptor orientations at 0°, 45°, and 90°
Fig. 3. (a) The microvilli orientation of microphthalmic cells on the ventral dorsal and ventral sides of the left eye in Stomatopods; (b) Four channel polarization distance model
Fig. 5. Calibration of solid/gas targets in infrared fused images with different channel polarization distances
Fig. 6. Image of solid/gas heat source processing results (The yellow box area represents the target of interest (high-temperature iron sheet), and the blue box area represents the selected background)
Fig. 7. Results of different channel polarization distance combinations
Fig. 8. The detection results of the solid/gas heat source part of the image testing set with different channel polarization distance combinations
Table 1. Objective evaluation data on the processing results of polarization distance models in solid/gas heat source scenarios
View tableView in ArticleTable 1. Objective evaluation data on the processing results of polarization distance models in solid/gas heat source scenarios
Different models GSC SCR FD BSF PD2 0.0805 1.1280 0.0063 1.8236 PD3 0.5487 1.4710 1.4348 0.0928 PD4 0.4208 1.3787 0.8741 0.0937
Table 2. Objective evaluation data of polarization distance combination results for different channels
View tableView in ArticleTable 2. Objective evaluation data of polarization distance combination results for different channels
Fusion channel GSC SCR FD BSF 234 0.4646 1.4133 1.0636 0.1309 223 0.1246 0.7851 0.0224 0.6238 443 0.4331 1.3918 0.9332 0.0938
Table 3. Detection results of different channel polarization distance combination datasets
View tableView in ArticleTable 3. Detection results of different channel polarization distance combination datasets
Datasets mAP0.5 mAP0.5:0.95 234 99.3% 73.7% 223 99.4% 75.2% 443 99.2% 67.3%
Tools
Get Citation
Copy Citation Text
Shuai YUAN, Mingzhao OUYANG, Yuegang FU, Chennan YANG, Junwei XU, Qingqing CHENG, Yahong LI. Research on biomimetic polarization vision solid/gas heat source identification technology based on deep learning[J]. Infrared and Laser Engineering, 2024, 53(12): 20240324
Paper Information
Category: 图像处理
Received: Jul. 16, 2024
Accepted: --
Published Online: Jan. 16, 2025
The Author Email:
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20