在這篇論文中，我們提出一個可以在行動裝置上辨識手上硬幣位置與大小的系統。此系統利用行動裝置的攝影機取像，將圖像的色彩空間從 RGB 轉換為 Lab，利用特徵保留分群法量化轉換後的圖像，接著將圖像內容的邊線找出來，並且利用此邊線資訊偵測出圓，最後計算出硬幣的位置與尺寸大小。由實驗結果顯示，此系統能夠成功將圖像上的硬幣偵測出來。

In this manuscript, a system which automatically measures the number of coins by using mobile devices is
proposed. The proposed system converts the color space of captured image from RGB to Lab, applies feature-
preserving clustering algorithm to quantize the transformed image plane, finds edges and circles in quantized
image, and fi nally determines the number of coins. The experimental results show that the proposed system can
successfully detect coins in the given images.