利用視訊偵測果蠅睡眠系統之研發

Abstract

一般研究果蠅睡眠行為實驗，皆為將果蠅置於ㄧ細玻璃管內，利用紅外線偵測器置於管子中間，當果蠅通過中間時才判定移動，檢查果蠅是否在單位時間內移動到特定地點。傳統方法屬於被動偵測模式，若果蠅長時間置於管子某一側，則可能會造成誤判成睡眠，所以可能會忽略一些細部資料以及無法偵測區域。

本論文提出方法解決傳統實驗的問題，在此使用攝影機並利用近紅外線(波長850nm)當夜間照明，截取連續的果蠅影像，利用影像處理分辨出果蠅目前所在位置，比較前後影像資訊，判別出果蠅是否正在睡眠，本論文能提供更完整的果蠅睡眠行為資料，也提供果蠅活動行為資料等等。

A traditional experiment setup for studying sleeping pattern of fly was to place a fly in a glass tube. An infrared motion detector was placed in the middle of the tube. It detected fly motion if a fly moved across the middle, and the fly was considered awake. The conventional experiment is a passive approach. It loses information if a fly moves only on one side of the glass tube. In this thesis, we present an experiment setup that solves the problem occurred in the passive approach. Furthermore, we can design experiment that studies the impacts of social activities to the sleeping pattern. We used near infrared LED (wave length 850 nm) as the illumination light source which can be visible by a video camera that has infrared filter removed. A computer image processing was applied to compute the location of the flies. By comparing the previous and current locations of the fly, we derive the motion information of the fly. Our experimental results show that the proposed method can replace the convention method.