Region of Interest Detection in Thermal Image of DC Motor Using Image Processing for Average Temperature Calculation

Anomalies in temperature, such as excessive heat, often indicate potential equipment breakdown or malfunction. Therefore, it is crucial to implement preventive maintenance programs that closely monitor these variations. In motor thermal modelling, calculations involving the convection thermal coefficient are necessary. The coefficient relies on the average temperature of the motor surface, traditionally measured using a thermocouple. Alternatively, thermal imaging technology provides a non-contact and non-intrusive approach. However, using the predefined shapes provided by Testo IRSoft Software may include irrelevant areas, which can affect the accuracy of the average temperature calculation. To address this limitation, this paper proposes thresholding segmentation, an image-processing technique to segment only the relevant area of interest. By utilising this technique, the average temperature can be calculated more accurately. As a result, the average temperature of the segmented Region of Interest (ROI) is higher than the fixed shape ROI (IRSoft Tool). The average temperature difference is between 1.06 and 11.34%.