Illustrating traffic accident based on two-view 3D position

• Main Authors: Chang-Yi Lin, 林長毅
• Other Authors: Wen-Kai Tai
• Format: Others
• Published: 2013
• Online Access: http://ndltd.ncl.edu.tw/handle/22604358207072767019

碩士 === 國立東華大學 === 資訊工程學系 === 101 === In this paper, we propose a system for the assistant of traffic accidents handling procedure. The proposed system analyses the photographs of the traffic accident scene to extract the evidences of it; hence, the traffic police can handle the accidents efficiently, and prevent traffic congestion. The precise information of the nearby roads is obtained through Google Maps by the GPS information in the photograph. We first apply Speeded Up Robust Features (SURF) to extract the corresponded feature points from two photographs taken at two different views. Then, we calculate the scale-related 3D position of these feature points in Euclidean space according to the perspective projection by the related distances among the feature points in the photograph. We also combine Google Maps to locate the accident scene, and obtained the information of nearby roads by the GPS information in the photograph. Finally, user can adjust the reconstructed 3D traffic accident scene in the scene editor by the obtained position of the feature points, and then export the traffic accident graphs. The contributions of the proposed system are as follow: First, we take advantage of the information in the photograph to reconstruct the 3D scene of traffic accidents, making the handling of traffic accidents more efficient. Second, with the roads information obtained through Google Maps, the exported graph is much more precise than the manually drawing one.