基于分类器的图像模糊边缘检测快速算法  

Fast fuzzy edge detection algorithm based on classifier

作　　者：鲁继文[1] 张二虎[1] 

机构地区：[1]西安理工大学信息科学系,陕西西安710048

出　　处：《计算机应用》2005年第10期2374-2375,2380,共3页journal of Computer Applications

摘　　要：通过对Pal-K ing边缘检测算法的分析,提出了一种新的模糊边缘检测快速算法。首先对图像进行模糊增强,然后依据当前像素及其8-邻域像素的灰度,设计了一个分类器,通过计算相对于该分类器的模糊隶属度函数值,对像素进行边缘分类;最后锐化所得的边缘像素,剔除噪声。算法抛弃了Pal-K ing方法中复杂的迭代运算,同时也克服了Pal-K ing算法中对图像低灰度值边缘信息的丢失,还可以通过设置不同的参数来检测不同细节的边缘。实验结果表明,该快速算法比Pal-K ing算法的边缘检测能力更强,同时运算速度提高了约20倍。A new fast fuzzy edge-detection algorithm was presented by analyzing and improving Pal-King fuzzy edge detection method. Firstly, this algorithm enhanced the original image in fuzzy way and designed one classifier according to its and its 8-neighbor gray levels; secondly, classified each pixel into the proper edge class by computing and comparing its greatest fuzzy truth of membership; lastly, thinned the edge pixels detected and omit the noise edge pixels. This fast algorithm not only abandoned the complex iterative calculation of Pal-King method, but also conquered missing edge information which is low gray level and can detect various details by setting various parameters. Finally experimental results showed that this method improves the Pal-King algorithm efficiently and its calculation speed was about 20 times faster than Pal-King algorithm.

关 键 词：边缘检测 模糊分类器 隶属函数 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]