The segmentation of moving object in video sequences is attracting more and
more attention because of its important role in various camera video applications, such as video
surveillance, traffic monitoring, people tracking. and so on. Conventional segmentation algorithms
can be divided into two classes. One class is based on spatial homogeneity, which results in the
promising output. However, the computation is too complex and heavy to be unsuitable to
real-time applications. The other class utilizes change detection as the segmentation standard to
extract the moving object. Typical approaches include frame difference, background subtraction
and optical flow. A novel algorithm based on adaptive symmetrical difference and background
subtraction is proposed. Firstly, the moving object mask is detected through the adaptive
symmetrical difference, and the contour of the mask is extracted. And then, the adaptive
background subtraction is carried out in the acquired region to extract the accurate moving object.
Morphological operation and shadow cancellation are adopted to refine the result. Experimental
results show that the algorithm is robust and effective in improving the segmentation accuracy.