In order to solve the problem of low accuracy of existing gait recognition, a method of gait recognition based on multi-scale entropy and DTW (dynamic time warping) algorithm is proposed. The new method for gait recognition adopts the camera and the high sampling rate sensor, and the acceleration sensor of the mobile phone is adopted to collect the data. The sensor works at the low sampling rate, and the normal walking acceleration data of 50 volunteers are collected and processed by multi-scale entropy, from which the entropy values at various scales are obtained. Finally, the DTW algorithm is used to match the multi-scale entropy. The simulation results show that the method based on the combination of multi-scale entropy and DTW improves the accuracy of identity recognition, and the EER(Equal Error Rate) is 13.7%, which provides a new idea for gait recognition.