- 會議論文
- OpenAccess
一個使用標記控制分水嶺與平均移動影像分割方法於自動害蟲計數之研究
摘要
番茄是世界上重要且具有高經濟的農作物，也是台灣重要的農產品之一，雖利用設施栽培以避免害蟲發生，但仍無法阻隔小型害蟲之危害。因粉蝨體型細小且偏好聚集棲息於葉背，不易掌握蟲發生數量，經常因此錯失防治時機，或因盲目施藥，衍生農藥殘留超標，造成農產品安全問題。為達到粉蝨的低農藥有效管理，藉由週期性於設施內放置黃色黏紙陷阱來監測粉蝨之發生密度，當平均誘集粉蝨數量達防治門檻，即啟動防治工作。過去都是以人力來收集黏紙和計數害蟲數目，但此方式費時又費工，人力容易受生理和心理的因素，而造成計數錯誤。利用照像設備及網路的應用，收集每個黃色黏紙的影像，使用影像處理來達到自動害蟲計數的功能。因此本篇研究提出一個有效地去除黃色黏紙黏膠並保留粉蝨輪廓的梯度階層最小法，再使用標記控制分水嶺與平均移動影像分割方法，可將細小的害蟲和黃色黏紙分離取出，分割結果使用形態學來修補害蟲邊緣便可取得害蟲數目。實驗結果顯示所提方法效果良好。
並列摘要
Pest is one of the critical factors causing economic loss in greenhouse. The Integrated Pest Management has been widely applied to the agricultural practices to achieve minimizing crop damage, environmental contamination, and economic loss concurrently. One of prerequisites for the Integrated Pest Management that is to accurately investigate population densities of pest species. Yellow sticky traps are a common method for monitoring many pests in the greenhouse. Especially with the small size pests, such as greenhouse insect pests, the efficiency of human counting is low and unreliable depending on observation conditions of observers (e.g., identification skill, fatigue). Therefore, implementation of automatic pest counting is vital to the modern agricultural production. In the paper, we propose an automatic pest counting method. A procedure of the yellow sticky removed and segmentation was devised to count small size pests. Initially, the L_0 gradient minimization was used to remove the yellow sticky from background image. Marker controlled watershed and mean shift segmentation was used to segment insects. Finally, we could count pest using morphology. Experimental results demonstrate the effectiveness of the proposed method.