The walk-run sport is a movement that is very easy to implement and can help protect health and mental wellbeing. This article focuses on the gait biometrics characteristics in the process of walking, by studying the biometrics characteristics of the gait to provide data basis for the movement guidance and clinical therapy. Human gait is a biometric characteristics, with difficult to disguise, little affected by environment and other unique advantages. Along with the development of physiological signal analysis techniques, this paper studies the gait of eight different study objects in the walk-run movement course, and uses wavelet multi-scale entropy analysis method. Research shows that the mean values in the axial direction of 8 study objects’ peak values are larger and all concentrated between 3.0-3.6, with little individual differences; Conducts feature points information analysis of the two states of three study objects, the peak position overall moves in the low-frequency section in the non-normal state, the total entropy value increases in the Y axial direction and decreases in the Z axial direction. Through this theoretical explanation and the data statistical results, this research method can well solve the needs of motion analysis and medical data.