Research On Scene Recognition And Tracking Registration Technology Of Mobile Augmented Reality

Augmented reality(AR)is a technology of great concern in recent years.At present,the carrier of AR is usually smart phones.In this paper,we focus on the analysis of digital image processing algorithms from the research of the most basic level of AR.We have carried out theoretical research and comparative experiments on the feature detection algorithm,selected and improved the ORB algorithm,and proved the effectiveness of the improvement.And the feature matching algorithm was studied theoretically and compared with many experiments.Finally,the combination strategy of KNN algorithm,Lowe's algorithm and RANSAC algorithm was selected.Next,based on above research,the system framework was developed,including camera parameters,feature detection,feature matching,rendering and other system interfaces.Finally,the system was migrated to the mobile with scan mode and AR mode.It was proved that this mobile AR system basically meets the requirements of fluency,robustness,stability and so on.