Gabor-Fast ICA Feature Extraction for Thermal Face Recognition Using Linear Kernel Support Vector Machine

In this paper a framework is presented to deals with various aspects of face recognition like illumination, rotation and scaling. The proposed framework consists of three parts. In the first part Gabor filter is used over the thermal faces at different scales, locations, and orientations. In second part, the fixed point algorithm Fast ICA have been used over the Gabor filtered images to represent the images from higher to lower dimensional space for dimension reduction. Linear kernel support vector machine (LK-SVM), has been used for classifying the facial images. The thermal face images of IRIS Thermal/Visual Face Database have been used for experiment purpose and result shows that the proposed system is responding well over other techniques.