标题:
曲率估计及其在曲面检测中的应用Curvature Estimation Methods and Its Application in Surface Detection
作者:
邵晓芳, 彭志刚
关键字:
曲率, 曲率估计, 取向Curvature, Curvature Estimation, Orientation
期刊名称:
《Computer Science and Application》, Vol.5 No.6, 2015-06-29
摘要:
曲线或曲面的曲率信息是图像处理和计算机视觉的许多应用领域均需提取的重要信息，因而曲率估计成为底层处理的基本任务之一。在对原始曲率、高斯和平均曲率，基于圆的离散曲率，基于抛物线的离散曲率、基于Gauss-Bonnet理论的算法和基于Euler理论的算法等曲率计算方法进行基本描述的基础上，将现有的曲率估计方法进行了分类和总结，并通过实验验证了加入曲率估计可有效提高曲面检测方法的抗噪性。Curvature extraction is required for many applications in image processing and computer vision. Therefore, curvature estimation is a basic task of these applications. This paper gives a classification and summary for existing curvature estimation methods to facilitate further investigations based on describing the original mathematical curvature, Gauss curvature, circle-based discrete curvature, parabola-based curvature, Gauss-Bonnet based curvature, Euler-based curvature etc. Experimental results show that curvature information can improve the robustness to noise in surface detection.