An Adaptive Time Window Step Counting Method Based on Peak Detection

Abstract

The invention discloses an adaptive time window step counting method based on peak detection. A method for counting steps in an adaptive time window based on peak detection in the present invention, comprising: step 1. filtering and smoothing the acceleration data by using low-pass filtering and segmental smoothing filtering; step 2. changing the data processed by step 1 There are still some false peaks; therefore, there is at least one peak and one valley in each step cycle. Using this feature, all candidate peaks and peaks in this step are detected by calculating the maximum and minimum values. Valley value; Step 3. Filter the false peak value and false valley value through an adaptive time window for the candidate peak value and valley value detected in step 2. Beneficial effects of the present invention: the average accuracy rate of the algorithm proposed by the present invention is 98.2%, it has accurate step counting effect for different walking states and different postures, and it has better robustness for different users.

Description

technical field [0001] The invention relates to the field of adaptive time, in particular to a method for counting steps in an adaptive time window based on peak detection. Background technique [0002] Pedestrian dead reckoning is one of the more popular indoor positioning technologies at present. It does not rely on other devices to locate the user in real time when the initial position is given. Pedestrian dead reckoning uses the built-in inertial sensor readings of the smartphone to calculate the user's heading and step length, and estimate the position in real time. However, obtaining the user's real-time and accurate step count is a challenge for pedestrian dead reckoning. At present, the pedometer algorithms based on smart phones mainly include zero-crossing method, finite state machine method, correlation analysis method, and peak detection method. [0003] The currently proposed method of counting steps based on peak detection may have the problem of counting too m...

AI Technical Summary

Problems solved by technology

[0009] CN106289309A) The method includes: determining the peak point and the valley point and the number of the peak point and the number of the valley point according to the detection result, and taking the peak point or the number of the valley point as the step frequency feature, Determine the movement type of the person according to the stride frequency characteristics, and use the preset time window and the preset peak-valley window corresponding to the movement type. There are problems in this method: the preset time window and the peak-valley window Poor adaptability; different windows should be considered for multiple poses

[0012] 1. A fixed-size sliding window is easy to detect false peaks, which brings a lot of wrong judgments and poor adaptability

[0013] 2. The existing technology will detect the user's walking state before counting steps, which will affect the real-time performance of step counting

Images