Comparison of thresholding techniques for extraction of electrical hotspots from infrared images

Abstract

The objective of this study is to evaluate the performance of different thresholding techniques to extract the hotspot of electrical equipment’s through IRT imaging. Abnormal thermal pattern is showed by faulty electrical equipment’s compared to the regular and fault free components in the same condition. Detecting hotpots is very important to protect the equipment from compete failure or fire hazard. In production line sudden fault and shutdown causes huge loss. That’s why using thermal imaging can warn us before sudden fault by giving heat signature of the equipment. For getting the correct hotspot area from IRT image, different thresholding method is used by many researchers and technical person. However, for implementing automatic inspection of electrical equipment’s using thresholding techniques it is important to know which of the particular thresholding technique will perform best. For this reason, in this study seven different thresholding technique was applied to 24 different sample images of electrical equipment to measure the overall performance of the seven well known thresholding techniques. So, the result from this study will help in future to implement automatic thresholding-based condition monitoring of electrical equipment’s to detect any fault criteria without stopping the equipment’s or disassembling them.