Intelligent Gesture Recognition System Based On 3D Accelerometer

Acceleration-based gesture recognition is a very important aspect of human-computer interaction. Its main purpose is to using machine learning method offline training3D accelerometer data which collected by gravity accelerometer and reaching the target of online gesture recognition. It is of utmost importance in designing and implementing an intelligent and efficient human-computer application, such as online gravity sense games. So the primary task of implementing gesture recognition is to define a dozens of user gestures and make promise that the gestures have obvious distinctions. Then collecting dozens of data with everv gesture and extracting features from the collected data to make classification training. Finally, achieving the goal of online gesture data collection and identification.Gesture feature extraction is the key point of gesture recognition. The quality of feature extraction determines whether we can enhance the success rate of gesture recognition. From gesture proposed till now. a lot of gesture recognition algorithms have been developed, some of them have reached the satisfactory recognition effect. But there is still a problem in all algorithms where the calculation is too large. This problem limits the using scenarios of gesture recognition in human-computer interaction.In order to improve the speed of gesture recognition, we proposed Quick Online Gesture Recognition Algorithm. User can make normal and quick gesture recognition under the promise of the accuracv of identification.The algorithm extracts the information which reflects the meaning of user activities through analyzing the collected data. Therefore, compared with traditional methods, through extracting the time domain data as feature data and reducing the complex transform, our method is about two or three times quicker with the same recognition effect.Our algorithm also supports the online user gestures training, User can define custom gestures and train gesture data online. The custom gesture will be added to the gesture library which can enhance the interaction of human-computer interaction.