Modeling And Algorithm Realization Of Somatosensory Interaction Combines With Depth Sensor

With the rapid development of new technology of virtual reality and intelligent interaction, somatosensory interactive technology has become a research hotspot in the field of human-computer interaction.Somatosensory interaction instead of the traditional control method using the mouse and keyboard in a graphical user interface,its natural interaction mechanism makes the system have a better perception of the user’s intent.At present, somatosensory interaction techniques show a broad development prospects in the fields of medical rehabilitation, education and training, game development and industrial production.The researchers at home and abroad in the somatosensory interaction have some achievements, Nintendo, Microsoft, SONY and other technology companies have also launched their own somatosensory interactive products, but whether in theoretical research or practical scheme, many pressing problems still exist in somatosensory interactive technology:it is difficult to customize the sense of mission based body and to distinguish the rough motion and fine motion, lacking of the overall control scheme of common operating system, and other intelligent platform interfaces are not perfect and so on.To solve these problems, this paper proposes SPG state jump method and a CBCS gesture control strategy based on algorithm model COAM/BOAM, they can be used for modeling and realization of the most basic somatosensory task. Put forward a OFT finger feature detection method that combined with the bone data from depth sensor to solve the difficult problem of distinguishing rough somatosensory interaction action and fine action.Present a detection method of DBLM hand dynamic and TEF dynamic point filtering method in the operating system of somatosensory control,realize the accurate control of the operating system interface.For somatosensory interaction with other intelligent interactive platform, the humanoid robot imitation as an example,we propose a SPAC method to calculate the joint angle,with the humanoid robot characteristics, we construct a set of human motion imitation system, embodies the somatosensory interaction application in intelligent robot platform.Through controlling music playing to verify the CBCS somatosensory control strategy, control accuracy rate reached 98.3%, so this method exhibits good effect. In different gestures and light conditions, OFT finger recognition rate can reach more than 98.5%. DBLM hand dynamic method can not only shield the error caused the difference of users’ bones, and it has the adaptive function for different screen resolution,TEF filtering algorithm enables to improve the efficiency of 69.6% in subtractive dithering of dynamic points.Robot motion imitation in SPAC joint angle calculation method, the average recognition rate is 97.29%, the average response time is 33.94ms.Therefore, the algorithm proposed in this paper has good application value in somatosensory interactive technology.