一种基于模糊推理系统的边缘检测方法  

Edge Detection Based on Fuzzy Inference System

作　　者：陈宏希[1] 陈鹏[1] 

机构地区：[1]兰州石化职业技术学院电子电气工程系,甘肃兰州730060

出　　处：《自动化与仪器仪表》2016年第4期223-224,共2页Automation & Instrumentation

基　　金：2010年度甘肃省属高校基本科研业务费专项资金资助"往复式压缩机状态监测及RCM技术研究"(甘财教[2010]176)

摘　　要：提出一种基于模糊推理系统的边缘检测方法。此模糊推理系统选用Mamdani模型,使用基于规则的推理,系统输入是高斯函数一阶导数滤波基础上的像素梯度值和原始图像像素的标准方差,系统输出是图像像素被判定为边缘像素点的概率,输出概率经阈值判定,即可得到边缘图像。仿真结果表明,此方法相比传统Sobel等算子有一定的优势,能检测出更多、更加细致的图像边缘特征。An edge detection method is presented now,which is based on fuzzy inference system( FIS). The FIS's type is Mamdani,and the inference of this FIS is based on the rules,the two inputs of this FIS are the gradient of filtered image by Gaussian first derivative and the standard deviation of original image,also the only output of this FIS is the probability,which the pixel is determined as the edge with a threshold value,then the edge image is obtained. Experimental results show that the proposed method is more superiority than the traditional edge detection operators,such as Sobel and so on,the proposed method can detect much more edge pixels and the detail features of the image.

关 键 词：边缘检测 模糊推理 梯度 标准方差 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]