提出了一种Freeman链码与B样条曲线误差控制相结合实现轮廓拟合的算法，首先利用Freeman链码法进行边界跟踪，根据相邻像素点间的不同的链码变化关系，排除伪特征点，提取出轮廓中绝大多数特征点，然后结合基于误差控制的B样条曲线法，取得能够精确表示轮廓信息的特征点。本文算法即避免了使用曲率来进行求取特征点的复杂计算，提高了特征点检测速度，又提取出能够精确拟合轮廓的局部支撑点，实现了基于误差控制的轮廓曲线拟合。实验结果证明了本文算法的正确性。%This paper presents a contour fitting algorithm which combine Freeman chain code and error control of B-spline curves. Firstly, it uses Freeman chain code boundary tracking method, based on the difference between adjacent pixels variation of chain code, eliminate false feature point, the vast majority of the extracted contour feature points, and then combined with error control based B-spline curve method, to obtain feature points which can accurate represent contour information . This method of avoiding the use of strike curvature characteristic points for complex calculations to improve the feature point detection speed, and the extracted contour to precisely fit the local support points, error control based on the realization of the profile curve fitting. Experimental results show that the algorithm is correct.
展开▼
