室內藝術畫展覽空間之擴增實境導覽

Abstract

我們許多人都曾經有個共同經驗，那就是在大型的藝術展覽空間，像是博物館、畫廊或者是特殊的藝術展覽空間中，尋找想要觀賞的畫作之時，對環境感到困惑或者是迷路。 為了能提供遊客導覽上的協助，我們提出了一個基於擴增實境方式的導覽系統，幫助遊客找到想觀賞的畫作，並以擴增實境的方式透過使用者行動裝置的螢幕，提供更詳細的畫作內容介紹。我們所提出的系統主要分為兩個部分，畫作賞析以及擴增實境式的畫作搜尋與導覽。 為了提供遊客關於畫作精細的介紹，我們提出了一個基於SURF(speeded-up robust feature)比對的方法，能用以辨識使用者站在畫作前任何位置、利用行動裝置所拍攝影像中所含的畫作。我們用來辨識使用者從任何角度與位置所拍攝影像內之畫作的方法，是去比對使用者輸入之影像與一個特製的畫作資料庫的SURF特徵子。在特製畫作資料庫中的影像是利用一個基礎影像透過自動化產生不同視角影像的演算法所建構出來的。畫作的辨識結果包含畫作的標題、畫作懸掛的位置，以及遊客所輸入之影像是站在該畫作的哪一個朝向所拍攝的。 為了搜尋畫作的目的，一開始我們先使用SURF比對的方法將使用者附近之畫作與畫作資料庫做比對，以定出使用者目前的朝向及位置。接著使用Dijkstra演算法找出使用者所選取的目標畫作與目前位置間的最短路徑。然後，基於行人定位推算(pedestrian dead reckoning: PDR)的精神去設計一個導引演算法，透過訪問路徑上一個接著一個的畫作，引導遊客抵達目標畫作的位置。導引資訊以一箭頭形狀呈現，是透過讀取電子羅盤的數值以及目標畫作位置所產生的，並且即時擴增該箭頭在使用者行動裝置的螢幕上，以導引使用者沿著規畫好的路徑，朝著下一個畫作或所尋找的目標畫作前進。 本論文研究也提出良好的實驗結果，證實所提出系統的可行性。

Many people have the experience of getting confused or even lost in a large artwork exhibition space, like a museum, a gallery, or a special exhibition site, when they want to find a certain painting for appreciation. To provide such a guiding assistance for the visitor, an augmented reality (AR)-based guidance system is proposed in this study to help the visitor to search a painting he/she wants to appreciate and to explain the delicate information of the painting in an AR way in the visitor’s mobile-device screen. The proposed system includes two major components, namely, painting appreciation and AR-based guidance for painting search. In order to give the visitors an exquisite explanation about the painting, an SURF-based method is proposed to recognize the painting image which the user acquires from any orientation with respect to the painting with a mobile device. The way we recognize the user-taken painting image from any orientation is based on SURF-matching between the input image acquired by the user’s mobile device and a pre-learned painting database. The images in the painting database are created from a base image of each painting through an automatic algorithm for generation of perspectively-transformed image versions of the base image from any orientation. The recognition result includes the painting title, the painting location, the orientation from which the painting image was taken. And for the purpose of painting search, the SURF-matching method is used at first to find the user’s position and orientation by matching an image of a painting nearby against the painting database. Next, the Dijkstra algorithm is employed to find a shortest path from the user’s current position to a user-selected target painting. Then, based on the spirit of pedestrian dead reckoning (PDR), a guidance algorithm is designed, which leads the user to the target painting by guiding the user to visit one by one the series of paintings along the path. The guidance information in the form of an arrow is generated by use of the reading value of the electronic compass and the target painting location, and is augmented on the right time on the user’s mobile-device screen to direct the user to walk toward the next painting along the path or toward the target painting he/she is looking for. Good experimental results showing the feasibility of the proposed system and methods are also included.