For space large flexible structures, it is difficult to measure their vibration displacements under disturbances due to low frequency and large amplitude vibration characteristics. In this paper, a machine vision-based vibration displacement measurement method is presented to face this challenge and used to identify structural dynamics properties. First, the problems of low-frequency noise amplifying during the integration of acceleration to obtain displacement are analyzed. Then, the principles of the machine vision-based vibration displacement measurement are introduced. Experiments are performed to verify the accuracy and effectiveness of the method. In those experiments, a commercial digital camera is used as a vision sensor. Digital image correlation method is used to extract pixel displacement of the target points. The sub-pixel positioning technique is applied to improve measurement accuracy. The transformation relationship between physical coordinates and pixel coordinates is built by camera calibration. The vision-based measurement results are compared to the results obtained by linear variable differential transformer. The resulted structural dynamic properties are compared to the results given by acceleration measurements. The results show that the vision-based method can capture the primary dynamic properties of large flexible structures and give more accurate measurement results of vibration displacements.