Abstract: In order to overcome the phenomena of illumination changes, occlusion and fake target,
which appear easily in the moving target detection process, a random image selection and adaptive
background updating method for moving object detection is proposed in this paper. First, the
background is initialized through selecting an image from the video sequence randomly, and then it is
updated adaptively and iteratively based on change marker matrix to extract reliable background images.
The experiment results demonstrate that the background extracted through this algorithm does not have
the mixing phenomenon, so a reliable background can be constructed, and the target object can be
detected clearly even in the case of large changes in illumination and occlusion between moving objects.

Key words: moving object detection, random image selection, iteratively updating, change marker
matrix