In order to obtain gait features of human foot in normal walking process. This paper presents a fast and effective system for gait features acquisition. Feature points were attached to the specific parts of human lower limbs and a video camera was used to acquire images of a walking process. In addition, axis values of feature points can be acquired by image binarization processing. At the same time, foot trajectory of fixed center of mass and gait speed were abstracted based on the least square method (LSM). Finally, the gait of lower limb rehabilitation robot was planned according to the gait features, abstracting the gait trajectory of rehabilitation robot ,Moreover, gait speed of different ages were analyzed. Experiments results show that the system is feasible and stable, which provides a powerful data basis and authentication methods for the analysis of movement and control of lower limb rehabilitation robot.