A Picture is Worth a Thousand Words: Share Your Real-Time View on the Road

Abstract

The visual information provided by applications facilitating urban communications such as Google Street View and Waze scene report is intuitionistic and useful since it conforms to human cognitive habits. However, current street view services cannot provide real-time information, and the scene report service is passive. An on-demand and real-time visual-information-providing mechanism is still unavailable. In this paper, we develop the crowdsourcing-based Real-time View Share (RVShare) system, which provides pictures of requested locations taken by travelers just passing by. To enable the RVShare system, we propose a view-sharing job distribution mechanism, where a wavelet-based vehicle prediction scheme and a tree-searching-based tracking scheme are developed to select vehicles for view sharing. We also design a simple but effective incentive mechanism to encourage more travelers to participate into the view-sharing activity, where the rewarding process takes the quality of uploaded pictures into account. Moreover, we develop a processing flow for uploaded pictures to improve the accuracy of the picture quality evaluation and verify the picture content. Comprehensive experiment results from road tests are conducted to evaluate the performance of the RVShare system.