基于关键点的轮廓描述方法

Representation method of outlines by using critical points

为了有效表达图像的轮廓,给出了一种新颖的关键点描述轮廓方法。首先采用一种快速的点弦累加距离技术(chord-to-point distance accumulation,CPDA)用来提取角点;然后用椭圆模型找到相邻角点间的平面曲线的极值点,这样可以更好地表达轮廓的结构,通过相邻点间的曲线拟合出图像的轮廓。通过理论的推导和实验结果验证了所提出的方法比前人的工作有了更准确的模拟。

To better represent the outlines of subjects in an image,this paper proposed a novel approach by using critical point. Firstly,this paper used a fast chord-to-point distance accumulation（ CPDA） to extract corner points. Then it used an ellipse approach to find the extreme point on the planar curve between adjacent corner points,which could be represented the contour structure better. Finally,it obtained contours in the image with a fitting algorithm processing on these points. According to the theory analysis and experimental results,the proposed method obtains higher precision of the representation than previous works.