摘要
传统的Canny边缘检测算法采用高斯滤波会导致图像过度光滑,在梯度幅值计算时仅考虑x方向和y方向会丢失掉重要的边缘信息.此外,采用双阈值法检测和连接边缘时易导致低强度边缘点和噪声点的混淆.针对传统Canny算法的不足,提出了融合数学形态学滤波技术改进的Canny边缘检测算法.融合数学形态学滤波技术对图像做滤波处理,在3×3窗口中心像素的4个方向确定像素点的梯度幅值,最后根据梯度方向的信息检测和连接边缘点.实验结果表明,该算法在对噪声有较高抑制能力的同时,边缘检测的准确性和完整性得到了有效提高.
The traditional Canny algorithm using Gaussian filter causes the image too smooth.The important edge information is lost because of only calculating the gradient amplitude of the x direction and the ydirection.In addition,confounding some low intensity edges and noise is induced when using the double threshold method to detect and connect the edges.In light of the disadvantages of the traditional Canny algorithm,an improved Canny edge detection algorithm fused with mathematical morphology technology is proposed.The mathematical morphology is applied to filter the image in this paper,and the gradient of the pixel is determined in four directions of the center of the 3×3 window.Finally,the edge detection and connection method is adopted based on gradient direction.The experimental results show that the proposed algorithm can suppress the interference of noise better.Both the accuracy and the integrity of the edge detection can be improved effectively.
作者
李东兴
高倩倩
张起
蔡亚南
吴秀东
LI Dong xing;GAO Qian qian;ZHANG Qi;CAI Ya nan;WU Xiu dong(School of Mechanical Engineering,Shandong University of Technology,Zibo 255019,China)
出处
《山东理工大学学报（自然科学版）》
CAS
2018年第6期1-5,共5页
Journal of Shandong University of Technology:Natural Science Edition
基金
国家自然科学基金青年基金项目(51705296)
山东省自然科学基金项目(ZR2014JL027
ZR2016EEM20)
关键词
传统Canny算法
边缘检测
数学形态学滤波技术
检测和连接
traditional Canny algorithm
edge detection
mathematical morphology technology
the edge detection and connection
