A moving object detection method based on Lucas-Kanade optical flow vectors compensation algorithm is proposed.It is used to detect and track the moving target and is especially effective when the target is sheltered and makes sharp motion.This method is realized by using Lucas-Kanade′s algorithm to calculate the optical flow vectors of frames,and then choosing the optical flow vectors of the image frames next to the lost frame to predict the position of the lost target and make a full image sequences.Results show that this algorithm can detect moving objects more accurately and reliably when compared with other traditional optical flow methods.