基于NURBS曲面拟合的图像边缘检测方法

Edge Detection Based on the NURBS Surface Fitting Method

提出了一种利用NURBS曲面拟合进行图像边缘检测的新方法。与Haralick对称区域正交多项式曲面小片拟合方法不同 ,NURBS曲面拟合克服了Haralick方法对图像区域划分和灰度平稳的限制 ,且不再需要窗口区域对称。利用NURBS曲面拟合构造简单、算法灵活、易于控制的特性 ,提高了曲面小片构成的灵活性与边缘检测的精度及计算效率。本文给出了NURBS曲面拟合基本小面构成、拼接及求导方法 ,并以一个真实图像进行了实验验证 ,结果证明 ,该方法稳定可靠 。

This paper proposes a novel algorithm for image edge detection.Different from Haralick's facet fitting method,this method is based on the NURBS surface fitting and overcomes the limitation of Haralick's method.It can be used to detect edges and filter the noise effectively and robustly.Experimental results obtained on the real world images by using this algorithm present better performances We stress that this approach can be used to other image analysis.