Abnormal action tracking using robot vision system

Abstract

Nowadays medical technology makes the inevitable trend of an ageing population. So intelligent home care is an important direction of development of science and technology, in particular, elderly in-home safety management issues become more and more important. In this research, we propose a low of operation algorithm and use the triangular pattern rule, then can quickly detect abnormal movements of humanoid by the installation of a robot with camera vision at home that will be able to judge the abnormal movements of in-home elderly in real time. In this research, we will present a preliminary design and experimental results of abnormal movements from body posture that utilizes image pre-processing and three triangular-mass-central points to extract the characteristics. The result shows that the proposed method would adopt the few characteristic value and the accuracy can reach to 90% a single character posture. Furthermore the accuracy can be up to 100% when a continuous-time sampling criterion and support vector machine (SVM) classifier are taken.