An efficient moving target identification method

Presents an efficient method which uses two neighboring frames in image sequences for target identification. Using statistical information about the background noise and candidate regions´ noise after background registration, we can determine those candidate regions that have the same or similar noise distributions to the background´s which should be background regions, and those candidate regions that have different noise distributions from the background´s which should be the target region. In particular, when there is only one target in the image, we can simply determine that the candidate region whose noise distribution is most different from the background´s is the true target