Image Classification of Substation Equipment Based on BOF Image Retrieval Algorithm

Qingsheng Zhao; Yuying Wang; Dingkang Liang; Zun Guo

doi:10.3788/LOP57.181011

Journals >Laser & Optoelectronics Progress >Volume 57 >Issue 18 >Page 181011 > Article

Laser & Optoelectronics Progress
Vol. 57, Issue 18, 181011 (2020)

Image Classification of Substation Equipment Based on BOF Image Retrieval Algorithm

Qingsheng Zhao^1、*, Yuying Wang¹, Dingkang Liang¹, and Zun Guo²

Author Affiliations

¹Shanxi Key Laboratory of Power System Operation and Control, College of Electrical and Power Engineering, Taiyuan University of Technology, Taiyuan, Shanxi 0 30024, China

²School of Electrical and Electronic Engineering, North China Electric Power University, Beijing 102206, China

show less

DOI: 10.3788/LOP57.181011 Cite this Article Set citation alerts

Qingsheng Zhao, Yuying Wang, Dingkang Liang, Zun Guo. Image Classification of Substation Equipment Based on BOF Image Retrieval Algorithm[J]. Laser & Optoelectronics Progress, 2020, 57(18): 181011 Copy Citation Text

show less

Fig. 1. Flow chart of our algorithm

Download full size

Fig. 2. Feature point extraction of SURF algorithm. (a) Isolation switch; (b) cable port

Download full size

Fig. 3. Infrared image and corresponding BOF feature image. (a) Transformer contact; (b) current transformer; (c) cable port; (d) isolation switch

Download full size

Fig. 4. Classification results of different KNN models. (a) Fine KNN; (b) Medium KNN; (c) Coarse KNN; (d) Cosine KNN; (e) Cubic KNN; (f) Weighted KNN

Download full size

Fig. 5. Classification results of images in the test set. (a) Correct; (b) wrong

Download full size

Algorithm	Feature point	Running time /s
SIFT	12364	1.8
SURF	7259	1.2

Table 1. Extraction efficiency of SIFT and SURF algorithms

Classifier	Prediction speed	Classification model setting
Classifier	Prediction speed	Distance standard	Number of neighboring samples
Fine KNN	medium	Euclidean distance	1
Medium KNN	medium	Euclidean distance	10
Coarse KNN	medium	Euclidean distance	100
Cosine KNN	medium	cosine distance	10
Cubic KNN	slow	cubic distance	10
Weighted KNN	medium	distance weight	10

Table 2. Classifier parameters

Classifier	Classification accuracy
Classifier	1	2	3	4
Fine KNN	94.18	91.18	92.12	90.12
Medium KNN	94.11	91.12	92.12	90.55
Coarse KNN	25.00	25.00	33.33	33.33
Cosine KNN	95.59	93.94	92.47	90.71
Cubic KNN	92.65	90.12	89.94	88.12
Weighted KNN	91.18	89.55	91.18	89.29

Table 3. Classification accuracy of different imagesunit: %

Classifier	Classification time
Classifier	1	2	3	4
Fine KNN	8.19	8.04	10.21	9.98
Medium KNN	8.24	8.09	10.04	9.91
Coarse KNN	9.26	9.09	10.67	10.55
Cosine KNN	7.71	7.67	9.21	9.11
Cubic KNN	10.36	10.18	11.52	11.39
Weighted KNN	6.04	5.95	9.86	9.71

Table 4. Classification time of different imagesunit: s

Qingsheng Zhao, Yuying Wang, Dingkang Liang, Zun Guo. Image Classification of Substation Equipment Based on BOF Image Retrieval Algorithm[J]. Laser & Optoelectronics Progress, 2020, 57(18): 181011

Download Citation

Set citation alerts for the article

Tools

Set citation alerts for the article

Save the article for my favorites

Paper Information