本論文設計了兩種輔助帕金森氏症病友的復健系統－「步態復健系統」、「復健動作分析系統」。帕金森氏症可以透過復健，減緩運動障礙病症的惡化速度。本論文的目標就是期許帕金森氏症病友，能透過這兩套系統的輔助，在家中進行自主復健，達到控制病情之目的。 帕金森氏症病友在行走的步態上，通常會產生「帕金森式步伐」(Parkinsonian Gait) ，指的是上半身前傾、小碎步、交叉步態、加速步伐、步行凍結等等的症狀，這些病症會嚴重影響病友的行動能力。「步態復健系統」利用擴增實境的概念，將虛擬視覺線索利用微型投影機投射在實際地面上，讓病友在行走的時候，能根據此虛擬視覺線索跨步前進。另外，藉由放置在病友腳部的距離感測模組，計算步伐數，並偵測小碎步與交叉步態是否發生，一旦出現任何一種步態異常，系統就會發出警示語音來提醒病友應注意的事項。本系統透過提供視覺與聽覺的線索，希望能輔助病友行走，避免產生帕金森式步伐。 帕金森氏症是一種進行性的動作障礙疾病，隨著時間的進展，病症逐漸加重。透過各種復健動作的實施，能有效減緩病友的動作障礙程度。「復健動作分析系統」，能將Kinect擷取到的人體骨架資料，取出特徵，輸入自我組織特徵映射圖進行訓練，找出各動作的代表軌跡。當病友使用本系統進行復健動作分析，系統就會利用本論文所提出之SOMMTS (The SOM-based Motion Trajectory Similarity Measure) 演算法，辨識其復健動作的種類，回饋動作正確性分數、分析圖表與姿勢增減建議，提供病友復健動作的改進方向，以確實實行各種復健動作，提升復健效果。This thesis presents two rehabilitation systems to assist Parkinson’s patients: a gait rehabilitation system and a rehabilitation movement analysis system. The deterioration rate of Parkinson’s disease can be slowed down by regular rehabilitation. The objective of this thesis is to assist patients for rehabilitation at home, and to avoid the disease getting worse. Parkinson's patients may develop 'Parkinsonian gait', which is characterized by small shuffling steps, crossing gaits, freezing of gait, and so on. These symptoms seriously affect patient's walking ability to get close with people and nature outside, and that would endanger the physical and mental health. Therefore, we hope to provide patients with visual and auditory aid by combining augmented reality and distance sensor system. Expect patients to gain willingness and convenience for rehabilitation, and recover their normal gait in the past. The proposed gait rehabilitation system uses the concept of augmented reality by projecting virtual visual cues on the ground, which can assist patients with these cues in ambulation. Besides, this system detects whether the height of steps are not high enough and the gait is cross or not. Once the violation is happened, this system will give the alarm to notice patients. As time goes on, the motor symptoms of Parkinson’s disease would worsen gradually. The implementation of various rehabilitation movements can effectively relieve these symptoms. The proposed rehabilitation movement analysis system uses Kinect to capture human skeleton from the rehabilitation movement, and finds out a representative trajectory of each type of movement. When the movement is detected, this system will analyze it through the proposed SOMMTS system. The accuracy and some reference charts will be given to encourage patients to do rehabilitation more actually.
