本論文植基於一個視覺監控之室內小丑魚養殖的自動投餌系統，系統內主要包含有喜好區域（Region of Interest, ROI）設定、小丑魚資料擷取及小丑魚自動投餌判斷。經由本論文所提出的方法將能達到小丑魚自動投餌的目的，大幅降低人力及時間成本；此外，經由本論文實驗模擬結果可得知，利用本論文所提出的方法約可達到89.7% 的投餌正確率。

This paper presents an automatic feeding system of indoor breeding clownfish based on visual surveillance. This system mainly consists of region of interest set, extraction of clownfish information, and determination of automatic feeding clownfish. Using the proposed method, the purpose of automatic feeding clownfish is obtained and the human loading and computational cost are greatly reduced. Experimental results show that the accuracy can be maintained at 89.7% using the proposed method.