行動裝置拍攝影片之街景圖重建與快速索引

Abstract

隨著硬體價格的不斷降低以及影像擷取技術的快速發展，智慧型手機、數位攝影機、數位相機等等可以產生、儲存、觀看影像資料的行動裝置可說是人手一台，幾乎每個人都成了多媒體資料的製造者與使用者。「行車記錄器」為近年來新興的行動攝影裝置，其主要用途是當事故發生之時，可以作為釐清肇事責任之有效證據，逐漸成為車輛必備裝置，因此，「行車記錄影片內容分析」實為刻不容緩之研究主題。由於駕駛行程中路上所遭遇的各種人事物皆會記錄於影片之中，所以當有交通事故發生之時，警政單位常會需要收集事發當日有經過該路段附近之行車記錄器影片以利調查之進行。然而，因資料量龐大，如未經整理，將需耗費大量人力與時間去檢視每部影片找出相關片段。 本論文提出一套基於行動裝置拍攝影片分析之街景圖重建與快速索引方法，設計一演算法來分析影片畫面用以估算行動裝置之移動速度，進一步，基於此估算之畫面移動速度來重建長寬比符合原景像之街景圖，並與影片內容建立連結索引，如此一來，使用者將不需要看完整部影片，而可以迅速地瀏覽街景圖，從中選取所要檢視的片段。相信這套系統將可以有效提升影片審視與危安事件調查之效率。另一方面，此街景圖重建與索引系統亦可應用於建構「個人行車旅途日誌」，旅途中經過的景點街道，一樣可以利用所重建之街景圖來做索引，使用者可以快速地找到自己所欲觀看之特定路程的影片。

With the rapid development and reduced cost of digital video capturing devices, driving video recorders (DVRs) begin to gain widespread popularity. What happens along the way can thus be recorded in videos. However, searching for a specific scene or event among such massive video collections is laborious and tedious. In this thesis, we propose a speed-adaptive street view image generation system using the general front-mounted DVRs, requiring no additional devices deployed. Visual summaries of the street scenes along the way can also be provided, allowing users to retrieve a video clip corresponding to a specific road section much quickly. A novel approach of an efficient algorithm for estimating the distance a pixel has moved between the consecutive frames is also proposed. A street view image can be generated with an appropriate aspect ratio without demanding a constant driving speed. Experiments on an extensive data set show that the proposed system can efficiently generate street view images under different lighting and weather conditions, demonstrating its feasibility.