Abstract: Condition monitoring of electrical equipments by infrared thermography is strongly dependent on the hotspot temperature. However, accuracy of hotspot detection is dependent not only on the image processing techniques for segmentation but is also dependent on the camera parameters (set during thermograph acquisition). Uncertainty in image processing techniques is reduced through multilevel contrast enhancement and improved active contour modeling. In order to remove uncertainty due to IR camera settings, mathematical relationships between the camera parameters and the error in temperature should be determined. In this study, three different techniques namely linear regression, back propagation network, curve fitting are used for deriving two dimensional equations. Though BPN and curve fitting provide non-linear relationship, curve fitting is favored as it results in accurate results and lesser computational complexity.