Dynamic Gesture Recognition Based On Skeletal Feature Point Tracking

At the beginning of the development of human civilization,gestures were a crucial information interactive medium.People conveyed their ideas through gestures,and the development of gestures accompanied the progress of civilization.Gesture recognition is a computer based on the different changes of gestures,identifying the messages the target wants to convey and the control instructions to be issued.According to the principle of the study,gesture recognition has many different methods of partitioning,and it is common to distinguish gesture recognition from gesture recognition and static gesture recognition based on movement continuity of gestures.Gesture recognition is a new technology which has a wide range of practical value.It give full play to their initiative and actively gestures of naturalness and creativity,can let the user in the absence of training,let the machine learn the user's intention,the exchange between persons involved with the machine.At present,researchers at home and abroad are devoted to the study of gesture recognition,and now gesture recognition has become a hot topic.This paper mainly studies the dynamic gesture recognition based on skeleton feature point tracking,the main idea is to use the image binarization of the finger width information and key feature points get gestures after skeleton extraction,to tracking of feature points,to complete the basic dynamic gesture recognition.By video camera to obtain gestural gestures and through image pretreatment,the skin color model based on YCrCb space target movement analysis of gesture segmentation,get complete gestures part image binarization,then improved axial skeleton extraction,get skeleton,found in the skeleton finger palm point and endpoint for feature point tracking and combined with the finger width information,use the same gesture feature points of the trajectory of similarity,rule to determine,through dynamic planning of DTW algorithm frame feature point matching,fisting,palm sliding and single index finger click finish dynamic gesture recognition.Based on the algorithm of this paper,dynamic gesture recognition system is designed,the feature point sequence of template gesture is obtained through the sample training,and the verification work of the algorithm is accomplished by using the test gesture video.Based on the analysis of experimental results,the recognition rate of dynamic gesture recognition algorithm based on skeletal feature point tracking is high and real-time,which can realize accurate and effective gesture recognition.