Calculation method of three-axis accelerator by counting step

With the improvement of the quality of modern life, more and more people begin to pay attention to their daily health exercise, as an effective monitoring means of recording and monitoring exercise, has been widely used in mobile terminal applications. But most of the implementation is to calculate the moving distance by the GPS signal, effective but in the indoor or no GPS signal equipment can not work, at the same time the GPS accuracy of the results of the interference is also relatively large, this paper proposes a new step method, that is, through the device accelerator to calculate the number of steps, Not supporting the GPS device can also work properly, can be used to coordinate with the GPS step, so that the application of the use of more diverse scenarios.

1. Understanding Model Features

At present, most of the equipment can be detected in all directions of the acceleration detector, for example, iOS devices, we use its three-axis accelerometer (X,y,z axis to represent the direction of the graph) characteristics to analyze. It is used to detect the acceleration change in three directions of human walking. &http://www.aliyun.com/zixun/aggregation/37954.html ">nbsp;

Users in the horizontal walking movement, vertical and forward two acceleration will appear periodically, as shown in the action of foot-foot, as the center of gravity up to the single foot, vertical acceleration is a positive increase trend, and then continue to move forward, the center of gravity down two feet to the bottom, acceleration in the opposite direction. The horizontal acceleration decreases at the end of the foot and increases at the time of stride.

Reflected into the chart, we can see in the walking motion, vertical and forward generated acceleration and time is roughly a sine curve, and at some point there is a peak, where the vertical acceleration change is the largest, through the peak of the trajectory of the detection and acceleration threshold decision, you can real-time calculation of the user movement of the step number, You can also further estimate the user's walking distance.

2. Step algorithm

Because the user in the movement may hand the device or put the device in the pocket, so the device placement direction is uncertain, for this we calculate three acceleration vector length, obtains a walking motion the sine curve trajectory.

The second step is peak detection, we have recorded the last vector length and the direction of motion, through the change of vector length, we can judge the direction of the current acceleration, and the last save the direction of acceleration comparison, if it is the opposite, that is just over the peak state, then enter the step logic to step, or discard. By accumulating the number of peaks, users can walk.

Finally, the handheld device will have some low amplitude and rapid twitch State, or we commonly known as hand shaking, or a prank users want to use a short time and repeatedly shake the device to simulate people walk, these interference data if not removed, will affect the accuracy of the steps, for this interference, We can filter by adding thresholds and step-frequency judgments to the detection.

The fastest running frequency of the human body is 5HZ, that is, the interval of two adjacent steps is at least 0.2 seconds, as shown in the figure, we set the timespan in the process of our filtering high-frequency noise, that is, the pace of the situation too fast. At the same time, we compare with the last acceleration size, set up a certain threshold value threshold to determine whether the movement is effective, effective exercise can be recorded step.

3. About the Pedometer expansion

The above is a measure of acceleration based on the realization of the principle of stepper, known walking and running pace of experience, then a slight improvement can become a ranging meter.

Through the three-axis acceleration we can know the user's motion state, in addition to the steps, we can also pass the curve of the accelerator to judge the user fell state, made an elderly and children fall detection automatic alarm.