碩士 / 銘傳大學 / 資訊工程學系碩士班 / 103 / Most of people nowadays are busy with fast-paced living ,pressure and lack of exercises and it may easily produce physical pain. Besides, some people accidently have physical disabilities by traffic accidents, injuries from work and even exercises, so that "disability" which the term was considered mostly in elderly patents before, now can be happened to everyone.
The definition of "disability" should be regardless of age and categories, As long as there is a functional problem in body and needs long-term care support, including those who has dementia, psychosis, developmental delay and all the physical and mental disorders. Usually, the disabled patient has to go to hospital to complete the rehabilitation. For the patient, the rehabilitation is both time consuming and boring. It will also spend much medical resources in hospitals. Thanks to the advances in interactive technologies of somatosensory, those who have physical problems can have new and much more convenient ways for rehabilitation. In the thesis, we use well-known StarCraft game as an example applied with Kinect somatosensory machine to design and develop an interactive platform which can help disabled people to complete rehabilitation. The proposed system can respond different commands to StarCraft game by checking the degree of similarity between the rehabilitation movement of disabled patient and the rehabilitation professional. It also can dynamically adjust the levels of game playing difficulties to motivate patient to have the willing to continue the game rehabilitation.
Moreover, the proposed system can record the skeleton movement information during the game rehabilitation. Its similarity-checking methods apply simple Euclidean distance and dynamic time warping algorithms to calculate the similarity score for playing StarCraft. Through the time require in Euclidean distance calculation is significantly less than DTW, but its accuracy is slightly lower than DTW, this thesis also compares the performance difference between movement coordinate and vector data types by using interpolation calibration and the vector angle calculation to get the similarity of rehabilitation movements and then give some weighted values to calculate the normalized scores. Moreover, this thesis further improve the dynamic time warping through cutting the rehabilitation time smaller time slots in which to compare the differences in movements and then to calculate the correct rehabilitation similarity score more quickly and effectively. After a series of experiments on proposed system, the comprehensive testing results demonstrate all the effectiveness of different proposed algorithms from different kinds of rehabilitation actions to learn the cost-effectiveness for disabled people using Kinect somatosensory game rehabilitation.
| Identifer | oai:union.ndltd.org:TW/103MCU05392003 |
| Date | January 2015 |
| Creators | Ching-Hsun Hsieh, 謝慶勳 |
| Contributors | Chia-Hui Wang, 王家輝 |
| Source Sets | National Digital Library of Theses and Dissertations in Taiwan |
| Language | zh-TW |
| Detected Language | English |
| Type | 學位論文 ; thesis |
| Format | 52 |
Page generated in 0.0016 seconds