Facial Recognition with unique feature extraction and matching of similarities from thermal infrared images is presented. The proposed work is an integrated approach that consolidates the extraction of consistent features through the use of morphological operators, registration using Image Registration tool and matching these features through newly developed similarity measures for authentication. The new approach is developed with a thermal signature template formed using four images which is to be taken at various instant of time ensured that it does not affects any biometric matching process based on consistent thermal features. Thirteen subjects were used for developing the algorithm in a thermal imaging system. The highly accurate results were obtained in the similarity measures for skeletonised signature and also for anisotropic ally diffused signature. The present work is implemented in MATLAB and the results clearly demonstrates the ability of the infrared system to extend its application to other thermal imaging based systems.