基于改进Canny算子的图像边缘检测方法  

Image Edge Det ection Method Based on Improved Canny Operator

作　　者：肖庆追 李捷 陈鹤峰[2] 魏彩娥 XIAO Qing-zhui;LI Jie;CHEN He-feng;WEI Cai-e(School of Internet Finance and Information Engineering,Guangdong University of Finance,Guangdong University of Finance,Guangzhou 510521,China;School of Applied Mathematics,Guangdong University of Technology,Guangzhou 510521,China;School of Entrepreneurship Education,Guangdong University of Finance,Guangzhou 510521,China)

机构地区：[1]广东金融学院互联网金融与信息工程学院,广东广州510521 [2]广东工业大学应用数学学院,广东广州510521 [3]广东金融学院创业教育学院,广东广州510521

出　　处：《数学的实践与认识》2022年第7期77-87,共11页Mathematics in Practice and Theory

基　　金：2019年度广东省高等教育教学研究和改革项目:“互联网金融背景下基于SPOC的翻转课堂模式在应用型高校的教学创新研究”(501190064)。

摘　　要：经典Canny图像边缘检测算法在面对复杂背景和椒盐噪声时会出现伪边缘或漏检等问题,影响后续图像分割,目标检测和识别.针对经典Canny算法高斯滤波和人工门限设置2个步骤进行优化改进,首先提出一种循环自适应滤波方法代替高斯滤波对图像进行平滑降噪,提升椒盐噪声抑制性能的同时较好的保留了图像中的细节信息,然后提出一种最小类内类间距准则的2-均值算法自动确定高低阈值门限,相对于人工门限设置方法具有更高的精确性和更强的适应性.基于标准图像库数据开展试验,结果表明所提方法可以明显提升经典Canny算法的椒盐噪声鲁棒性和复杂背景下的边缘检测性能.The classical canny image edge detection algorithm will have some problems such as false edge or missed detection when facing complex background and salt pepper noise,which will affect the subsequent image segmentation,target detection and recognition.In order to optimize the two steps of Gaussian filtering and artificial threshold setting of classical Canny algorithm,an iterative adaptive filter is proposed to reduce the image smooth and reduce noise instead of Gaussian filter,which improves the performance of pepper salt noise suppression,and retains the details of the image.Then,a 2-means algorithm with minimum class to class spacing criterion is proposed to automatically determine the high and low threshold The value threshold has higher accuracy and more adaptability compared with the manual threshold setting method.The results show that the proposed method can improve the robustness of salt pepper noise and edge detection,performance in complex background.

关 键 词：边缘检测 图像恢复 噪声抑制 复杂背景 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]