太極拳為一項結合平衡、強化、伸展的物理運動，曾被《時代雜誌》譽為最完 美的身心運動，也被應用於改善老年人或帕金森氏症病人的平衡能力與身體協調， 對心肺功能、柔軟度、平衡、張力與微循環的改善有相當大之益處，且對於心理及 精神的幫助也備受肯定，還能減緩認知功能衰退與提升免疫力，被視為是一種身心 運動。因此，有愈來愈多人開始學習太極拳，但學習者卻無法隨時都有專家的指導 與判定其動作之好壞。本研究將利用微軟Kinect感應器的紅外線攝影機，擷取人體 骨架關節位置之三維座標，再以三維座標計算出各關節與骨架組成的距離、角度與 比例，搭配專家對於各動作關鍵骨架角度之判定與評分標準，訂立出各特徵之容許 範圍與評量指標，開發一套太極八法電腦化自動評量系統，即時診斷學習者之太極 拳基本動作的優劣。比對電腦評量分類結果與專家評判結果，有60%的基本動作正 確率達到0.7 以上，也具有不錯的敏感度與特異度，故太極八法電腦化自動評量系 統能有效地區分出使用者在各太極基本動作的好壞。

Tai Chi, a physical activity integrated balance, strengthening, and stretching, has been called the “Perfect Exercise” by TIME magazine. It has been applied to improve balance in older adults or people with Parkinson’s disease. Moreover, Tai Chi movements are also helpful to mental health and spirituality, decreases the speed of the cognitive decline, and improves immune system. Hence, more and more people learn Tai Chi, now. However, most of them are learnt by watching video or films and without experts’ immediate instructions. It is hard to detect the goodness of Tai Chi basic movements by learners themselves. This study used Microsoft Kinect sensor to capture the three-dimensional skeleton points of human body and computed the distances between two joints, the angles between any two skeletons, and some proportion about the center of gravity. Based on these measurements and the tolerance ranges in specific areas determined by the expert experience, a computerized automatic assessment system for Eight Essentials of Tai Chi was proposed. A learner can find the weak movements immediately from the report of the system. According to the experimental results of comparing expert’s classification and the computerized classification, the developed assessment tool has good sensitivity, specificity, and classification accuracy. Therefore, the developed assessment can distinguish between good and poor basic movements efficiently.