Night-time pedestrian detection by infrared-visual video fusion
YU Le-bing · Computer Engineering and Applications Journal · 2009
An infrared-visual video fusion method is proposed for night-time pedestrian detection.At first,infrared video is denoised by spatial-temporal filtering and thermal-image feature of pedestrian,then bright region which corresponded to high temperature region in infrared video is segmented by seeded region growing algorithm,then targets are detected by region color and shape feature of pedestrian,at last,infrared video and visual video are fused together to provide more visual information in enhanced video.