透過您的圖書館登入
IP:3.21.104.72
  • 期刊

台灣地區天蠶蛾(鱗翅目:天蠶蛾科)網路影像辨識系統

The Web-based Image Recognition System for Giant Silkworm Moths (Lepidoptera: Saturniidae) in Taiwan

摘要


台灣鱗翅目中的天蠶蛾,擁有的美麗翅膀經常成為昆蟲愛好者的觀察目標。目前雖有不少網站提供天蠶蛾圖像供比對,但圖像比對的客觀性仍是一大困擾。本研究建置並提出一個天蠶蛾影像的自動辨識系統,使用者僅需透過網頁上傳天蠶蛾的影像,系統在自動進行去背、擷取物件正規化後,將對天蠶蛾影像特徵值進行辨識。由於天蠶蛾的翅膀有前後翅之分,本系統透過翅膀的形態數位分類特徵,能將上傳影像自動進行前後翅分類並進行資料庫之辨識比對;對於同時上傳前後翅影像的使用者,兩影像與資料庫影像進行辨識,可進一步提昇辨識率。國內天蠶蛾正式紀錄物種數量有16種,目前本系統已建立其中10種天蠶蛾的數位影像特徵分類資料庫,在前、後翅影像同時輸入之個體辨識率可達97%。若能擴充資料庫之個體數量、物種數量或其生態資料,將可提昇系統之辨識率與應用範圍。

並列摘要


The beautiful giant silkworm moth with large size is one of the most interested insects in Taiwan. Till now, the website-providing automatically retrieval system based on the uploaded wing images of giant silkworm moths or other insects doesn't been created. In this study, we develop a web-based image recognition system for giant silkworm moths. As the user uploads the wing image of moth, the system will extract wing image from background, normalize the image, and compute the contour and color characteristics of the wing pattern. In the recognition processes, the wing images will be classified into anterior or posterior wing automatically. Currently, 10 out of 16 species of giant silkworm moths in Taiwan are contained. Results show the accuracy will be more than 97% as both anterior and posterior wings images are provided. The recognition possibility would be enhanced when the data base of individuals, species, or ecological information are increasing.

被引用紀錄


王信夫(2012)。影像辨識系統在台灣常見蝴蝶種類辨識之應用〔碩士論文,國立臺灣大學〕。華藝線上圖書館。https://doi.org/10.6342/NTU.2012.10979

延伸閱讀