tracking is a dicult task in computer
vision, which is usually affected by color, surrounding
illumination, variation of the object's appearance and other
factors. In previous years, many algorithms can only set up
xed appearance models to track object. Recently, more and
more tracking algorithms have been proposed to deal with
object appearance variation and illumination change.
However, these algorithms are easily inuenced by the
background and can only track the object for a short time.
novel incremental principal component algorithm with
classier detection is proposed to solve the driing and
long-term tracking problems. Numerous experiments
demonstrate that the proposed algorithm is more robust
than several state-of-the-art algorithms.