Typical Physical Activity Recognition Algorithm And Application Study Based On Android Platform

This paper describes built-in sensors to automatically detect whether the smartphone is worn in a pocket and then only when the smartphone is detected in a pocket, physical activity recognition will be activated without any limitation of firm attachment. In contrast to the previous work, this paper intends to recognize the physical activities when the phone’s orientation and position are varying. By introducing a method to judge whether the phone is in a pocket, we investigated the data collected from six positions of seven subjects, chose two signals that are insensitive to orientation for classification. Decision trees (J48), Naive Bayes and Sequential minimal optimization (SMO) were employed to recognize five activities:static, walking, running, walking upstairs and walking downstairs. Compared the classification results of three classifiers, the results demonstrated that the J48classifier produced the best performance (average recognition accuracy:89.6%), and then we chose the J48classifier as online classifier.After identifying human movement state, we studied the meter step algorithm under walking and running. One way to detect step is to use smartphone, and another is to use the Miniholter developed by shenzhen institutes of advanced technology.Due to the fall phenomenon is common in the elderly, we studied the falls monitoring algorithm. We fixed the phone in the position of chest for data collection, and let volunteers simulate the elderly fall under the condition of the laboratory. Finally, we design an early fall warning system. We also collected the ECG data in human movement, and studied the ECG under different motion state, in order to find a useful way to help us to identify physical activity.The following three aspects are included in this study:1. Using smart phone built-in sensors to automatically detect whether the smart phone is worn in a pocket and then only when it is detected in a pocket, physical activity recognition will be activated without any limitation of firm attachment. Using WEKA data mining software for comparison of classification algorithm, and finally developed a real-time motion classification application. 2. Studied the fall identification method and designed a kind of fall early warning system, and combining the research of ECG auxiliary motion recognition of human movement.3. Developed the meter step algorithm, using the built-in motion sensors with smart phone’s positions and orientations varying.