Key points are not available for this paper at this time.
When text visualization comes into the field of MAR (Mobile Augmented Reality), some problems go with it. Firstly, the natural object in the real scene holds abundant text information. Trying to recognize the natural object as a visual element rather than a marker, make text data intuitively understand. But MAR has been restricted by mobile computing speed and capability. Until now, the limitation makes text visualization difficult to meet the users' increasing needs in NFT (Natural Feature Tracking). Tools have been needed to recognize natural object to obtain considerable text information; even worse, being limited by a small screen of mobile, two-dimensional text visualization cannot intuitively and comprehensively present all kinds of text information, and interaction methods in text visualization are too insufficient to meet users' needs. The main technical contribution to the work is a natural feature recognition algorithm. The work is based on WebAR (Web-based AR) and an inverse BOVW (Bag of View Words) model. Transforming the feature points matching problem into a tree search problem. Compared with conventional BOVW, inverse BOVW drastically reduces the consumption of space and time; WA VIS shortens the time cost of others ranged from 8.5% to 13.7% with accuracy in 95%. And also, WebGL (Web Graphics Library) enriches interaction methods of text visualization in three-dimensional space with MAR. The test results verified that the proposed method can effectively improve the recognition speed and accuracy of the natural image in MAR for text visualization, enrich interaction in text visualization with MAR.
Pei et al. (Fri,) studied this question.