| Recent advances in technology have produced imaging sensors sensitive to both visible and long-wave infrared (LWIR) light. These sensors produce images with the two channels in perfect coregistration. While such sensors have generated excitement in the electro-optical community, the exploration of their use in computer vision has been limited. An important question to the computer vision community is how such sensors might be helpful for main stream computer vision problems, along with thermal infrared sensors alone.; At the same time, the face detection and tracking problems have become increasingly important as a necessary preprocessing step for face recognition in biometric identification applications. While recent progress has been rapid, face detection is still sometimes the limiting factor in recognition performance.; This thesis will show that these two open problems are related in that new sensors can be a powerful tool in improving face detection and tracking performance. In addition, to improving detection and tracking performance by applying new sensors, this research improves the performance of existing algorithms on visible imagery. |
