介紹了一種基于機器視覺的大張鈔票離線檢測系統。分析了系統原理，點陣字符分割
算法及識別方法。提出了利用圖形邊緣模板匹配搜尋特征點自動提取ROI 區域，通過計算連通區域包含像素進一步去除噪聲的方法。系統實現了大張鈔票的實時檢測標廢。試驗表明識別速度為15 張/秒，廢品指示準確率為100%。
關鍵詞：大張鈔票；機器視覺；點陣字符；圖形邊緣匹配；實時
Abstract: A system based on machine vision for off-line checking the printing quality of bank note was proposed. The principle of the system, the segmentation algorithm of dot characters and recognisition method are introduced. Using Shape edge template matching technique to find the feature point and abstract the area of ROI automatically, calculate number of the pixels that the connected areas have to remove noise further. The system realize real-time detecting of banknotes. Experiments show the processing and recognisition speed is 15 papers per second, the instruction? accuracy is 100%.
Key words: bank note; machine vision; dot characters; shape edge matching; real-time