GMM (Gaussian Mixture Model) and HMM (Hidden Markov Model)-based step phase detection method

Abstract

The invention discloses a GMM (Gaussian Mixture Model) and HMM (Hidden Markov Model)-based step phase detection method, which relates to the field of pedestrian navigation and comprises rough division based on a Gaussian Mixture Model) and further probability estimation based on a Hidden Markov Model. According to the application requirements of the method, a gait cycle is divided into a support phase and a swing phase, according to the characteristic that the swing amplitude difference between the support phase and the swing phase is large, it is assumed that the support phase and the swing phase belong to two different Gaussian distributions, and the support phase and the swing phase are divided into two clusters through a Gaussian mixture model clustering method. A walking process is divided into a series of continuous gait time periods by analyzing a threshold condition of a rough division result of a Gaussian mixture model, and each gait time period is defined as an observation value, so that a group of observation sequences is obtained. The method not only improves the reliability and robustness of the existing gait detection method, but also enables the selection of the application method to be simpler, more convenient and more flexible.

Description

technical field [0001] The invention relates to the field of pedestrian navigation, in particular to a gait detection method based on GMM and HMM, which belongs to the gait analysis method based on inertial technology. Background technique [0002] Human movement is a highly complex dynamic process, but human gait is a regular and periodic process. In the field of pedestrian navigation, gait analysis is to detect gait phases in combination with the laws of human kinematics, and determine the support phase in the gait cycle. The pedestrian navigation algorithm can take advantage of the fact that the stride speed in the support phase is zero, and perform zero-speed correction in a timely manner, so as to suppress the accumulation of navigation position errors and improve the long-term navigation accuracy of the system. The existing gait detection methods all have the shortcomings of manual setting and individual adjustment of detection parameters. In the actual navigation pr...

AI Technical Summary

Problems solved by technology

Existing gait detection methods all have the shortcomings of manual setting and individual adjustment of detection parameters

In the actual navigation process, the detection results are often inaccurate due to unreasonable method design or inappropriate parameter settings, which in turn affects the correctness and effectiveness of the zero-speed correction method and introduces navigation errors of varying degrees.

Images