A Comparison of Color and Infrared Stereo Approaches to Pedestrian Detection

This paper presents an analysis of color and infrared stereo approaches to pedestrian detection. We design a four camera experimental testbed consisting of two color and two infrared cameras that allows for synchronous capture and direct frame-by-frame comparison of pedestrian detection approaches. We incorporate this four camera system in a test vehicle and conduct comparative experiments of stereo-based approaches to obstacle detection using color and infrared imagery. A detailed analysis of these experiments shows the robustness of both color and infrared stereo imagery to generate the dense stereo maps necessary for robust object detection and motivates investigation of color and infrared features that can be used to further classify detected obstacles into pedestrian regions. The complementary nature of color and infrared features gives rise to a discussion of a feature fusion techniques, including a cross-spectral stereo solution to pedestrian detection.