Research On Kinect Recognition Based On Kinect And Its Application In Teaching

With the development of interaction technology, the Human-Computer Interaction pattern has gradually be evolved from the traditional interaction pattern to body Interaction pattern which were based on action recognition. One of representatives of the interaction device is called 3D body tracing camera. In order to apply this new interactive model to the teaching system, and use body movements to operate the PPT, this paper carried out the following work:(1) Analysis the real demand about using the new device in our PPT teaching system, and make a survey about the frequency of teacher’ common posture in classroom. And this survey was ready for the selection of posture.(2) On the basis of 20 skeleton point model which was build before, the new model we build is based on 13 skeleton points. This new model is used to reduce the negative effects when skeleton points keep out each other. Throw the simplified model, huge needless skeleton data were deleted. The efficiency of recognition can also be increased.(3) Do two experiments in order to test the success rate about posture recognition. One is the static action, another is dynamic action recognition. These two experiments were used to test the simplified model’s success rate, real-time, and robustness. And the static posture recognition result will determine the optimum recognition distance to access.(4) The recognition algorithm was improved, new algorithm matching the multiple body parts’ position with key frame and extraction for the number of key frames. Based on experiments, we choose several actions to control the PPT in classroom.The work of this paper made an innovation to help the action recognition. And the method we used to recognize the action can also be used for reference. Our research laid a good foundation for the subsequent research.