摘要
为了解决传统Sobel算法在边缘提取时存在的定位不准确、边缘线条较粗、细节缺失的问题,提出了一种改进算法。首先使用增加方向模板后的Sobel算子对图像进行卷积,再对梯度图像进行边缘检测得到双线条图像,并以此作为掩膜来消除原梯度图中的伪边缘,从而对边缘线宽进行细化,然后使用Otsu算法将图像进行二值分割,最后通过数学形态中基于击中击不中变换的细化对图像进行全局处理来消除图像纹理干扰点。实验表明,与传统的边缘提取算法相比,改进后的算法定位准确,边缘较细,并保持了原图像的细节。
In order to solve the shortcomings of low edge position accuracy and coarser edge and detail deficiency of the traditional Sobel algorithm,an improved algorithm is proposed.Firstly,the Sobel operator which increases the direction template convolutes image,and then defects the edge of the gradient image to obtain the double line image,with the double line image as a mask to eliminate the false edge of the gradient image,thus refining the edge line width.The Otsu algorithm binaries the image and aims the global image based on the hit-miss morphological thinning to eliminate the image texture interference.The result shows that the improved algorithm has the higher location accuracy and keeps the detail of original image,achieving the thinner edge than traditional Sobel algorithm.
作者
陈恺煊
刘昕
王咚
CHEN Kaixuan;LIU Xin;WANG Dong(School of Printing,Packaging Engineering and Digital Media Technology,Xi’an University of Technology,Xi’an 710048,China)
出处
《西安理工大学学报》
CAS
北大核心
2018年第1期116-121,共6页
Journal of Xi'an University of Technology
基金
陕西省重大科技创新资助项目(2008ZKC02-13)
