Thermal Face recognition system plays a crucial role in area of pattern recognition and computer vision. Major problems affecting the accuracy of the thermal face recognition systems are environmental factors such as illumination changes and physiological changes created by the subjects. From the available literature, it is already proven that, local texture descriptors based approaches are predominantly impervious to temporal variations. In this paper, recently proposed Local binary pattern based feature descriptor such as Local Derivative Ternary Pattern (LDTP) is used for feature extraction. The accuracy of the recognition system also depends on the distance metric and classifiers. This paper is experimented with Support Vector Machine (SVM) classifier. Thermal face database such as UCH Thermal Temporal Face (UCH-TTF) and PUCV Thermal Temporal Face (PUCV-TTF) are used for experimental investigation.
Volume 11 | 01-Special Issue
Pages: 543-547