Abstract: Optical flow is one of the most important technologies for motion analysis,which can reconstruct the motion of objects from consecutive image frames and hence can be utilized for object detection,motion tracking and feature identification.Based on the principle of global optical flow,the main techniques are reviewed in detail from three sections:non-uniform illumination correction,energy function construction and optimization scheme design.Meanwhile,three main difficulties for global optical flow analysis are studied thoroughly,which include large displacement computation,occlusion detection and weak texture estimation.At last,challenges are summarized and prospects of this technique are made.