Protective Clothing Detection of Substation Workers Using S-HOG+C Operator

This paper focuses on the application of the protective clothing detection system by using S-HOG+C operator for substation workers in order to meet the practical needs in a complex environment. An image is firstly divided into three cells including a helmet, upper and lower part of a body according to the characteristics of objects. The HOG and HOC features of the three cells are extracted individually to train a classifier. A linear Support Vector Machine is used in our study. The S-HOG+C operator is used to improve the detection accuracy on the protective clothing of substation workers. Three evaluation indicators are used to analyze the performance of the S-HOG+C operator. The experimental results suggest that the S-HOG+C operator perform better than the HOG+C operator in the protective clothing detection for substation workers. This work is supported by Science and Technology Project of Guangdong Power Grid Co., Ltd. (GDKJXM20162351).