摘要
在质量检测、遥感图像处理和三维重建等很多计算机视觉应用场景中,准确检测图像中的边缘信息起着至关重要的作用。提出一种基于高斯拟合的亚像素边缘检测算法。算法首先利用Canny边缘检测算法得到像素级的边缘位置信息,再通过高斯拟合的方法将边缘位置准确度提升为亚像素级。与传统的亚像素边缘检测方法相比,提出的算法运算时间更少并且精确度更高。不同强度噪声干扰下合成图像的边缘检测效果验证了算法的鲁棒性和精确度(在10%的噪声下,边缘误差在0.006像素数量级)。另外,算法还在基于多角度真实图像的光滑物体三维重建系统中取得了较好应用。
In many computer vision applications such as quality examination,geospatial image processing and 3D reconstruction,the accurate detection of edge information in images is a critical step. We propose a sub-pixel edge detection method based on Gaussian fitting method. First,our method uses canny edge detection algorithm to obtain pixel-level edge information and then utilizes Gaussian fitting to refine the accuracy of edge information to sub-pixel level. Compared with traditional sub-pixel edge detection methods,the proposed algorithm not only achieves better detection accuracy but also costs less time. Experimental results using synthetic images corrupted by different levels of noise show the robustness and accuracy of the proposed algorithm（ with 10% image noise,the edge location error is at the 0. 006 pixel magnitude）. In addition,the algorithm is also well applied in the 3D reconstruction system of smooth objects based on multi angle real images.
作者
韩东
李煜祺
武彦辉
Han Dong;Li Yuqi;Wu Yanhui(The 28^th Research Institute, China Electronics Technology Group Corporation, Nanjing 210007, Jiangsu, China)
出处
《计算机应用与软件》
北大核心
2018年第6期210-213,229,共5页
Computer Applications and Software
关键词
亚像素边缘检测
高斯拟合
Sub-pixel edge detection
Gaussian fitting
