A real-time recognition algorithm of scenes on smartphones

Currently, the used SIFT and SURF algorithms cannot meet the demand of higher real-time identification applications, and these algorithms have a lot of problems, including a long matching time, a large amount of memory usage and computational complexity and so on. In this paper, we propose a method for real-time recognition on a smartphone, through shortening the time of feature point detection and reducing the complexity of feature point location on scale space to ensure real-time identification and accuracy. The experimental results show that this algorithm can effectively run on resource-constrained ordinary smartphone with good versatility. At the same time, it can achieve real-time recognition of the scene and consume less memory resources, so it is suitable for using in practical applications.