上海文通大学 Research Status of Human-Computer Interaction In the future HCI,emotion will be absolutely a factor that cannot be ignored.Therefore,the facial expression recognition is supposed to be improved greatly in order to occupy an important position. 5.4 Action Recognition 5.4.1 The recognition of human action Perhaps no one would deny that action recognition is one of the most common ways of human- computer interaction.Considering the advantages and convenience of human action,scientists have systematically researched the human action recognition technology,including the hand posture, upper body gesture recognition and gait recognition (Cao,2012). The detection of static hand gesture is a progress of analyzing different finger and palm postures in a certain moment.Usually there are two methods,the first one is based on models including the card model,the skeleton model and the distribution model.The second one,however,is based on hand shape characteristics and contour feature. In comparison of hand gesture,upper body can express more abundant and accurate information.A movement of one's upper body can cause the acceleration signal.The typical process of human movement patternrecognition is to collect acceleration signals,then after smoothing and de-noising, valid signals can be extracted to make up feature vectors and get training models(Xu,2013). The gait feature is an emerging behavior characteristics,mainly meaning the habitual way of walking.It can reflect user's psychological features and physical condition for the simple reason that there can be subtle difference when people are in different conditions.Gait recognitions usually contains four progressing,motion detection,cycle detection,feature extraction and recognition algorithm (Wang Hou,2007). 5.4.2 The application of action recognition Fig.8 Action recognition applied in somatosensory games Thanks to the continued research and development,systems based on action recognition have already been used in lots of fields.According to news,in June 2010,one Japan company introduced some household electrical appliances such as TV that can be controlled by gestures like waving or rotating hands.Then in 2013,Samsung S4 achieved to be operated without touching but only 11/19Research Status of Human-Computer Interaction 11 / 19 In the future HCI, emotion will be absolutely a factor that cannot be ignored. Therefore, the facial expression recognition is supposed to be improved greatly in order to occupy an important position. 5.4 Action Recognition 5.4.1 The recognition of human action Perhaps no one would deny that action recognition is one of the most common ways of human￾computer interaction. Considering the advantages and convenience of human action, scientists have systematically researched the human action recognition technology, including the hand posture, upper body gesture recognition and gait recognition (Cao, 2012). The detection of static hand gesture is a progress of analyzing different finger and palm postures in a certain moment. Usually there are two methods, the first one is based on models including the card model, the skeleton model and the distribution model. The second one, however, is based on hand shape characteristics and contour feature. In comparison of hand gesture, upper body can express more abundant and accurate information. A movement of one’s upper body can cause the acceleration signal. The typical process of human movement pattern recognition is to collect acceleration signals, then after smoothing and de-noising, valid signals can be extracted to make up feature vectors and get training models (Xu, 2013). The gait feature is an emerging behavior characteristics, mainly meaning the habitual way of walking. It can reflect user’s psychological features and physical condition for the simple reason that there can be subtle difference when people are in different conditions. Gait recognitions usually contains four progressing, motion detection, cycle detection, feature extraction and recognition algorithm (Wang & Hou, 2007). 5.4.2 The application of action recognition Fig. 8Action recognition applied in somatosensory games Thanks to the continued research and development, systems based on action recognition have already been used in lots of fields. According to news, in June 2010, one Japan company introduced some household electrical appliances such as TV that can be controlled by gestures like waving or rotating hands. Then in 2013, Samsung S4 achieved to be operated without touching but only