In HV substations, thermal effects of temperature and resulting heating losses have a profound impact on the life of substation equipment. In order to efficiently detect these irregular heating patterns occurring in substation, a non-intrusive infrared thermography (IRT) based technique is suggested for detecting these abnormal heat variations in the electrical .substation equipment. In order to analyze thermal images of the substation, manual analysis may consume a lot of time and efforts and also there is every possibility of human errors in detection of faults. Thus to automate process of thermography, image processing techniques have to be opted for. In this paper, the process of image processing methodology consisting of feature vector extraction and image classification based algorithm are proposed here. Initially, two of the intelligent image processing techniques i.e. feature vector extraction techniques, K-means clustering and Fuzzy C-means (FCM) clustering, are proposed here and their results will be compared with each other with respect to their performance and efficiency. Then the substation equipment condition will be further evaluated and classified using Support Vector Machine (SVM) classifier to classify thermal images into two categories as normal or abnormal. The programming of the above mentioned techniques is done using MATLAB software. The proposed classifier technique results indicate higher accuracy above 90%.