LAR: a low-power, high-precision mobile phone-based AR system
Xiaoming Dai,Fei Shang,Tianzhang Xing,Feng Chen,Baoying Liu
DOI: https://doi.org/10.1007/s00779-020-01421-3
2020-07-14
Personal and Ubiquitous Computing
Abstract:Mobile crowd sensing is a novel large-scale sensing pattern, which applies users' smart devices to analyze social context and human activity, and then learns the intelligent information serving for various innovative services. An augmented reality (AR) system can bring more authentic experience for human life, and more and more scholars are increasingly interested in this technology. However, there are some problems in the previous methods, such as the expensive deployment, low accuracy, and high latency. These problems greatly limit the application of augmented reality systems. In this paper, we design a lightweight augmented reality system, called LAR, which can recognize the target object quickly and precisely by using a feature matching algorithm. In LAR, the target object is shot twice to get the distance between the user and the object, and we take the distance as the feature of the target object. Furthermore, the feature extraction algorithm, named s-SURF, is designed to extract image features. LAR combines the distance with image feature and then matches with these images stored in database. In addition, the flutter-free algorithm is used to denoise and get a clearer image. Finally, we design a prototype system to evaluate LAR performance. The accuracy of LAR is 86%, and the time delay achieves 141 ms.
computer science, information systems,telecommunications