Research On Constrained Virtual Indoor Scene Synthesis Method

In virtual reality applications,users can experience virtual reality with high sense of reality and immersion through virtual scenes presented by head-mounted displays.Because the user’s entire field of vision is covered by the monitor,the real physical environment cannot be perceived.Therefore,the virtual scene displayed in the device needs to meet the specific geometric layout to avoid the collision of users during roaming.The color and texture of objects in the virtual scene also need to meet certain constraints for the user’s realistic and immersive needs.This paper studies the synthesis method of virtual indoor scene under specific constraints,so that users can safely carry out virtual reality roaming in the real environment,and ensure users’ high sense of immersion and realistic experience.The main innovations of this paper are as follows:1.A method of synthesizing virtual indoor scene layout under geometric constraints of real environment is proposed,which can avoid collision when users roam in real environment and realize interaction with real objects at the same time.This method detects the volume and position of interactive planes and objects from the 3D scanning of real environment,and abstracts the furniture object relationship and layout template from the indoor scene data set.Based on these two parts of information,the virtual indoor scene synthesis process is transformed into a combinatorial optimization problem.In this paper,simulated annealing algorithm is used to solve the problem and get the furniture layout.2.A method of automatic texture mapping for virtual indoor scenes based on theme color constraints is proposed,which can meet users’ requirement of realistic virtual environment and improve users’ sense of immersion.For an indoor scene containing multiple rooms,the user only needs to provide the theme color constraint of one room.This method predicts the theme color of other rooms based on the GAN network,and synthesizes texture maps for each room according to the predicted results.In order to enrich the final texture results,an interior furniture texture database containing 4336 texture images is also provided.