| Summary: | 碩士 === 國立中興大學 === 應用數學系所 === 103 === Obesity is one of the serious problem in the modern world, and pose a serious impact with human in the future. It can cause many deadly diseases, such as: heart disease, diabetes, etc. Physical activity is strongly important solution to reduce the threat of obesity. It is difficult to quantify the energy consumed of the activities, while the accelerometer is a kind of tool to tell the user the amount of energy consumed.
Type of activity, intensity and duration can affect the body''s energy consumption, so this research proposes an identification of distinguishing the gait of walking and up the stairs. All of the 20 young subjects that were age of 22.7 ± 2.1 years old, height 170.7 ± 6.5 centimeters and weight 67.9 ± 12.1 kilograms. They wear three-axis accelerometer on the left ankle, using the sampling frequency is 200Hz and the sensitivity is 68count/(m/s^2) to collect acceleration signal.
We found out that the acceleration signal of subjects were quite different in walking and walking up the stairs. In this research, we used nine statistical parameters (maximum, minimum, mean, standard deviation, range, sum, median, skewness and kurtosis), the threshold, using the relationship between the parameters entering perceptron to calculate decision boundary , then do the gait recognition. The accuracy of the results were over 90%. Therefore, this research proposed a simple and effective method of gait recognition.
Keywords：gait recognition、perceptron、decision boundary
|
|---|
