Thermal Images for Face Recognition Using Statistical Analysis

Despite successes in indoor access control applications, imaging in the visible spectrum demonstrates difficulties in recognizing the faces in varying illumination conditions. Face recognition using different imaging modalities, particularly infrared imaging sensors, has become an area of growing interest. The use of thermal infrared images can improve the performance of face recognition in uncontrolled illumination conditions. In this paper, we present a new technique for face recognition by calculating statistical parameters of componentbased thermal images. We propose a feature vector that consists of the following different feature parameters: the first moment, the second moment, and the thermal image histogram. The calculation of these parameters is implemented at the component level, beside the whole face image. The features used in the local analysis are less sensitive to illumination changes, easier for estimating the rotations, have less computational burden, and have the potential to achieve higher correct recognition rates. Hence, the proposed system exploits the advantages and the characteristics of: thermal images, component-based approach, and the statistical features. The experimental results reveal that the new system can achieve a success rate of 94.2% when implemented on the AIAOU Database. Keywords-thermal image; face recognition; histogram analysis; moments