中文摘要：

简要介绍了目标测头视觉坐标测量系统，分析了测量系统成像目标的图像特征。在此基础上，提出了一种识别和提取CCD图像中目标椭圆的新方法。该方法对CCD图像中的每一个封闭轮廓进行椭圆拟合，根据评价椭圆形状误差的线轮廓度大小来判定该轮廓是否为目标椭圆轮廓。实际检测结果表明，该算法原理正确，易于编程实现，能有效剔除噪音图像，准确提取目标椭圆，具有广泛的实用性。

英文摘要：

A vision 3D coordinate measuring system is introduced using a special probe, and the image feature including target of the probe is analyzed. A new algorithm is proposed to recognize and extract target ellipses from a CCD image. Ellipse fitting is applied to each closed contour in image captured when large or tiny invalid contours are excluded. And whether this contour is a target ellipse or not is judged by comparing its elliptical profile error with a fixed threshold, for it often is known to evaluate the shape error of an ellipse by computing its elliptical profile error. This algorithm is correct, efficient and easy to be programmed. The result of extracting test is obtained under multiple noise condition, and demonstrates that this new algorithm can eliminate the noise contour efficiently, and distill the target ellipses precisely.