基于边缘邻域梯度的椭圆参数估计算法  

Elliptical Parameter Estimation Algorithm Base on Neighborhood of Elliptical Edge

作　　者：秦大辉[1] 毛婷[1] 刘建军[1] 

机构地区：[1]西南石油大学土木工程与建筑学院,成都610500

出　　处：《煤矿机械》2011年第10期60-62,共3页Coal Mine Machinery

基　　金：国家自然科学基金资助(51074137)

摘　　要：研究了一种基于边缘邻域内的像素梯度椭圆参数估计算法。首先利用梯度幅度直方图和类内方差将椭圆边缘环状邻域提取出来;然后将环状邻域上所有的点代入椭圆二次曲线模型中并以其梯度值作为权值,通过最优化算法将椭圆参数计算出来。与传统的椭圆拟合方法进行对比,实验的结果显示新算法有很强的抗噪声能力和较高的精度,像素误差在0.03个像素左右。The ellipse＇s parameter estimation algorithm based on the gradient of the pixels in the edge neighborhood is presented.Above all,the ring neighborhood of the ellipse＇s edge is extracted based on gradient histogram and variance within clusters.Then,fit the ellipse conic curve in its gradient value as weights which the points of the neighborhood is used to calculate the parameters of the ellipse via the optimization algorithm.The proposed algorithm has been incorporated into a copyright-reserved a hand-held laser 3D measurement system.Comparison with other tradition approaches,it shows strong anti-noise capability and preferable accuracy in the real situation experiments.The error of pixel is about 0.03.

关 键 词：相机标定 特征提取 近景摄影测量 椭圆拟合 机器视觉 

分 类 号：TG315[金属学及工艺—金属压力加工]