Arkit-based Augmented Reality Spatial Interaction Design And Related Technologies

In recent years,with the improvement of mobile devices computing power and the upgrading of mobile user experience,a large number of mobile phone applications have built richer interactive experience through various augmented reality functions in ARKit platform.However,most of the current mobile augmented reality applications realize object recognition,three-dimensional positioning and real-time tracking by registering the target information in advance.Because the pre-registration method needs to obtain the information of the target in advance,it has great limitations for the application scenario,and the recognition process does not have generalization.In addition,most of the applications in interaction mode are still simply reusing touch screen gesture interaction mode.There is still a lack of relevant research on the layout and interaction mode of the interface in real space,which limits the development of spatial interface interaction mode.Through depth learning in computer vision and other related technologies,this paper realizes target recognition and location without pre-registration,explores the spatial interface layout and interaction rules under this mode,and tests their interaction performance.The main contents of this paper are the following three parts.1.The research of three-dimensional positioning without pre-registration.The network model with high speed and accuracy on mobile devices is obtained through theoretical research and experimental test.Then,the point clouds generated by ARKit are mapped to screen coordinates,and the three-dimensional point clouds information of the target is obtained by combining the two-dimensional checkboxes obtained from target detection.Next,through the distance distribution between the point clouds and the mobile phone,the noise point clouds generated by the background and the occlusion in the selection frame are detected,and the thr-ee-dimensional coordinates of the target are calculated after filtering.Finally,the Deep-Sort algorithm is extended to three-dimensional coordinates to achieve real-time tracking of multiple targets.2.Spatial interface layout and interaction research.Firstly,various spatial layout parameters(interface shape,interface distance,interface size)are obtained thr-ough pre-experiment.Then,the influence relationship between them and various user interaction experience factors(mobile change rate,mobile terminal mapping size,etc.)is theoretically studied,and the optimal model of various spatial layout parameters is analyzed.Next,through the experimental platform,the optimal distribution values of each user's layout parameters are measured and counted.Finally,the correlation between parameters is tested by statistical analysis and the spatial interface layout model suitable for most users is obtained.3.Based on the above technology foundation and the experimental results of the layout model,an interactive application example of information browsing based on ARKit to enhance the real space interface,AR Skim,is constructed.The performance and subjective satisfaction of users in three kinds of information browsing tasks are recorded thr-ough the application,and compared with the performance and subjective satisfaction of gesture interaction on traditional two-dimensional screen,it is proved that the performance of spatial interface in large-scale information browsing tasks is significantly better than that of traditional interactive mode.This paper starting from the interaction of spatial interface in mobile ARKit augmented reality environment,this paper studies the interaction arrangement rules and related technologies,and provides technical basis and interaction model for the research of unregistered target recognition and augmented reality information interface in ARKit environment.