Abstract –Object tracking is always a challenging research
to the computer vision community. It becomes more
difficult at night video systems due to low contrast against
the background. This paper is proposing a framework that
detects object and tracks it at low contrast night
surveillance video. A robust intensity statistics based
detection method has been designed for processing low
contrast frame and detect object structure from it. Based
on successful detection, it tracks the object using Kalman
filter algorithm.