應用LabVIEW於外來果實蠅即時通報系統之設計與實現

Abstract

由於交通工具的發展，使得現今社會的交流與貿易變得更加地頻繁。我國亦自國外進口多種水果，頻繁的貿易活動亦提高農產品挾帶外來種果實蠅入侵的機會。單純依賴進出口的檢疫工作，並無法完全防治外來種果實蠅的入侵。因此，防檢局在臺灣本島佈建了多套果實蠅誘捕裝置，並透過人工的方式定時定點的調查是否誘捕到外來果實蠅。此調查系統不僅耗費大量的人力，且無法即時通報防疫系統，減少外來種果實蠅所造成的潛在農業經濟傷害。 因此，本研究欲結合數位影像處理與自動化鑑別方式，偵測外來害蟲入侵事件；並輔以通訊技術，達到即時通報之功效，亦可使我國防疫與檢疫系統得以更加健全。 本研究透過LabVIEW程式建構了一套外來種果實蠅即時通報系統。本系統包含兩部分，分別為前端果實蠅誘捕與影像擷取系統，以及後端果實蠅影像鑑別與通報系統。前端果實蠅誘捕與影像擷取系統採用甲基丁香油誘引劑吸引果實蠅，透過網路攝影機擷取果實蠅影像後，以3G網路將影像經檔案傳輸協定回傳至後端伺服器。擷取後的影像透過本研究所提出之自動化影像鑑別演算法，運算出相似度，並篩選去除大多數的東方果實蠅影像後，即通報防疫檢驗人員以肉眼檢驗疑似外來種果實蠅的影像。 透過本研究所設計與實現的外來果實蠅即時通報系統所回傳的即時監測影像，可改善了原本需要大量人力資源的影像檢驗工作，加速外來果實蠅的檢疫與通報的流程，藉以減輕外來果實蠅造成臺灣農業上的經濟傷害。

Due to the rapid development of transportation, there is a growth in international trade. Taiwan imports many kinds of fruit, and frequent trading activities also increase the invasion opportunities of foreign fruit fly brought by imported fruit and other agricultural products. The invasion of foreign fruit fly, however, cannot be fully prevented by solely relying on import and export quarantine. Thus, the Bureau of Animal and Plant Health Inspection and Quarantine in Taiwan has set up fruit fly trapping devices to catch flies, and manually inspect the traps to investigate whether foreign fruit flies exist. This survey system is not only labor consuming but also not able to immediately provide the fly information to pest control programs to reduce the potential damage to agriculture caused by the foreign fruit fly. Therefore, this study combines digital image processing with automated identification to detect foreign pest invasion. The study also uses the communication technology to acquire real time reports. Such a design improves our epidemic prevention and quarantine system. In this study a foreign fruit fly real-time reporting system was developed by the LabVIEW programming language. The system consists of two subsystems. The front-end subsystem traps fruit flies and captures the images of the flies, while the back-end subsystem identifies different types of fruit flies and reports to a pest control center when anomalies are detected. In the front-end subsystem, fruit flies are trapped by methyl eugenol, and their images are taken by a webcam module. The images are then sent to the back-end server through a 3G network using the File Transfer Protocol (FTP). The back-end server compares the captured image to the templates in the database using an automated image identification algorithm proposed in this study. If an image does not match the templates in the database, the server declares the occurrence of foreign fruit flies, and automatically informs this situation to the pest control center. Through this research, a real-time foreign fruit fly image capturing and alert system is developed. This system greatly reduces not only the requirement on human resources for image inspection, but also the time needed between foreign fruit fly reporting and quarantine. With the above mentioned advantages, the proposed system is able to reduce the economic harm caused by foreign fruit flies in Taiwan.