Research On Virtual Rendering Technology Based On Depth Sensor

In recent years,due to the emergence of cheap sensors such as Kinect,researchers are increasingly enthusiastic about the development of virtual reality technology.Virtual reality technology can present a virtual reality environment where users can use the virtual environment for immersive experience or enhance their perception of reality.Due to the rapid development of virtual reality technology,the scope of application in real life has become wider.Due to the rise of online shopping,virtual fittings have become one of the most virtual reality products for users with its convenient dressing,friendly UI interface and various choices.In addition,in the presentation of virtual reality,the speed and authenticity of 3D reconstruction is also crucial.The point cloud of the input multi-frame scene or object is reconstructed by using the registration and surface method to reconstruct the model of the scene or object.The amount of data is huge,and the actual surface extraction of the object is difficult.This paper has mainly studied in the following two aspects.Firstly,for the virtual fitting mirror in the shopping mall,there is a strong sense of substitution,and the fitting room needs to make manual measurement through physical contact.This paper proposes a real-time virtual modeling technology based on virtual reality and rendering technology.Human body data is acquired in real time through Kinect sensors.Aiming at the inaccurate measurement of human parameters by single acquisition data,this paper proposes a method to combine the depth and bone data measurement parameters based on computer image processing technology.Experiments show that the height accuracy of this method can reach 97.54%.The method also has a significant advantage in stability.Then based on the Unity3 D modeling technology,a system is established to generate a human body model in real time.Secondly,aiming at the problem that the human dummy model in the home virtual fitting room is not real enough,this paper proposes a Kinect-based human body modeling method based on 3D reconstruction technology and point cloud filtering technology.This method uses Kinect to collect point cloud and preprocesse and register the point cloud.Aiming at the problem that the cumulative deviation of point cloud increases gradually during the registration process,this paper proposes a novel side registration method based on point cloud registration technology.The method starts registration from the side to the two directions.Side registration is utilized to reduce the cumulative deviation of the registration.Finally,based on the 3D surface reconstruction technology,the human body model generated by the registered point cloud is used.The experiment shows that the accuracy of the human body model constructed in this paper can reach 96.48%.