A study of the identification of overlapping coins by using the image processing technology

• Main Authors: Kai-Chun Hsu, 徐凱軍
• Other Authors: Kuang-Tsan Lin
• Format: Others
• Language: zh-TW
• Published: 2010
• Online Access: http://ndltd.ncl.edu.tw/handle/18953083296063115801

碩士 === 聖約翰科技大學 === 自動化及機電整合研究所 === 98 === The purpose of this thesis was to utilize the image processing technology and apply it to coin detection. This was primarily meant to improve the situation of 「unable to identify the overlapping coins」. Coins of different monetary value have different sizes and colors. Basically, these two features could be utilized as the classification basis. They could solve the drawback caused by the loss of coin image due to overlapping, as well as the reduction of the recognition rate. Besides, the problem of the rotation of coins would not be an issue. Therefore, much time could be saved. Furthermore, this method could provide the data entry of multiple coins. First, take the photo and engage in image processing. Then, utilize the Hough Transform to detect the circular shape. Afterwards, use the HSV to do color transformation on the photo in order to reduce the color difference between the coins. Then, collect the color information of the individual coins, respectively cut out the overlapping coins, and compare with the data bank samples. Therefore, by using the sizes and colors of the coins, the problem of 「not being able to identify the coins that overlapped more than 30%」 could be efficiently improved.