圖像的快速亞像素邊緣檢測方法

A Fast Sub-pixel Edge Detection Method for Images

劉力雙(Li-Shuang Liu)；張銚(Yao Zhang)；盧慧卿(Hui-Qing Lu)；趙琳(Lin Zhao)；王寶光(Bao-Guang Wang)

圖像處理 ； 亞像素 ； 邊緣檢測 ； Sobel算子 ； 曲線擬合 ； image processing ； sub-pixel ； edge detection ； sobel operator ； curve fitting

光電子·激光

16卷8期（2005 / 08 / 15）

993 - 996

簡體中文

提出了一種新型的測量圖像快速亞像素邊緣檢測方法。首先，利用標準的Sobel算子進行邊緣點的粗定位，確定邊緣點的像素級精度位置和邊緣的方向；然後，沿邊緣點的邊緣方向拓展像素，得到長度為6的像素灰度值向量，將向量帶入利用最小二乘曲線擬合方法得出的公式，求出邊緣點的精確位置，從而能夠實現亞像素邊緣定位精度。實驗證明：該方法的定位精度為0.l pixels，算法的運行時間為0.53 s。

A novel fast sub-pixel edge detection method for image measurement is proposed. Firstly, in order to find the edge point and edge direction at pixel level, the approximate positioning of the edge point is carried out by means of Sobel operator; secondly, the pixel is expanded along the edge direction of edge point to obtain the pixel grey vector with a 6 length. Then the accurate position of edge point can be extracted with a sub-pixel precision through the least square fit method. The experimental results show that the positioning accuracy amounts to 0.1 pixels and the running time of algorithm is 0.53 s.