Evaluation of video distortions on fused infrared and visible videos in surveillance applications.

The increasing level of uncertain security in the last years in our world, along with the availability of cheaper and more intelligent digital cameras technology is encouraging great interest in the development of video-based systems capable of observing and interpreting specific scenes, with the aim of detecting anomalies or events that may affect the safety, security, economics, and vital parts of human activity. Fusion of long wave infrared (LWIR) and visible videos is an approach to solve the disadvantages of using visible videos, which are sensitive to illumination variations. Video quality distortions in LWIR and visible videos are obstacles that limit the effectiveness on video surveillance tasks such as detection and tracking of cars and humans in fused LWIR and visible video sequences. However, the analysis about the impact of IR and visible video degradations on object tracking has received very limited study. But, how do video degradations affect LWIR and visible video fusion? Under what conditions of video distortion current object tracking methods in fused and visible videos can successfully work? How do video degradations affect the task of humans detection and tracking in videos resulting from LWIR and visible video fusion? The main objective of this proposal, which is oriented to answer these questions, is to evaluate the impact of LWIR and visible video distortions on fused LWIR and visible videos and human tracking performance in surveillance applications.