Thermal Imaging Detection Method of Leak Gas Clouds Based on Support Vector Machine

Jing Weng; Pan Yuan; Minghe Wang; Li Li; Weiqi Jin; Wei Cao; Bingcai Sun

doi:10.3788/AOS202242.0911002

Acta Optica Sinica, Volume. 42, Issue 9, 0911002(2022)

Thermal Imaging Detection Method of Leak Gas Clouds Based on Support Vector Machine

Jing Weng¹, Pan Yuan¹, Minghe Wang¹, Li Li^1、*, Weiqi Jin^1,2, Wei Cao², and Bingcai Sun³

Author Affiliations

¹MoE Key Lab of Photoelectronic Imaging Technology and System, Beijing Institute of Technology, Beijing 100081, China

²Beijing Wisdom Sharing Technical Co., Ltd., Beijing 100125, China

³CNPC Research Institute of Safety & Environment Technology, Beijing 102206, China;

show less

Abstract Get PDF(in Chinese)

Figures & Tables(9)

Fig. 1. Schematic diagram of SVM

Download full size

Fig. 2. Wide band gas infrared imaging system

Download full size

Fig. 3. Original images and difference images of different infrared video and their rendering pictures. (a)(e) ith frame; (b)(f) (i+5)th frame; (c)(g) difference images of ith frame and (i+5)th frame; (d)(h) color rendering of difference results on original image

Download full size

Fig. 4. Samples of positive and negative datasets. (a)--(c) Positive sample images; (d)--(f) negative sample images, which are people, trees, and pumping units

Download full size

Fig. 5. Optimization results for parameters C and γ by grid search method

Download full size

Fig. 6. Detection results of test set. (a) Positive sample; (b) negative sample

Download full size

Table 1. Test results of SIFT-SVM algorithm
View table
Table 1. Test results of SIFT-SVM algorithm
Parameter Accuracy /% Errorrate /% Precision /% Recall /%
Value 92.50 7.50 91.00 93.81

Table 2. Test results of HOG-SVM algorithm
View table
Table 2. Test results of HOG-SVM algorithm
Parameter Accuracy /% Errorrate /% Precision /% Recall /%
Value 82.50 17.50 86.00 80.37

Table 3. Performance comparison of two algorithms
View table
Table 3. Performance comparison of two algorithms
Algorithm Single data size /kB Iterations Training time /s Number of support vectors
SIFT-SVM 2--6 176 3.6165 326
HOG-SVM 41--63 824 137.7023 502

Tools

Get Citation

Copy Citation Text

Jing Weng, Pan Yuan, Minghe Wang, Li Li, Weiqi Jin, Wei Cao, Bingcai Sun. Thermal Imaging Detection Method of Leak Gas Clouds Based on Support Vector Machine[J]. Acta Optica Sinica, 2022, 42(9): 0911002

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites