改进免疫遗传算法的零值绝缘子分割方法研究  

Improved immune genetic algorithm for zero insulator imaging segmentation method

作　　者：钱艳萍[1] 何洪英[1] 蒋正龙[2] 赵纯[2] 关石磊 夏德分 

机构地区：[1]湖南大学电气与信息工程学院,长沙410082 [2]湖南省电力公司试验研究院,长沙410007 [3]湖南湖大华龙电气与信息技术有限公司,长沙410082

出　　处：《计算机工程与应用》2011年第31期157-159,170,共4页Computer Engineering and Applications

基　　金：国家重点产业振兴和技术改造项目(发改投资[2010]2272号);湖南省电力公司项目(2009年);中央高校教育基金项目(2009-2012)

摘　　要：根据零值绝缘子红外图像对比度差、目标边缘模糊、噪声大且目标温升比环境温升略高的特点,采用自适应中值滤波和均值滤波去除原始图片的噪声,采用对数变换增强图片的对比度;提出一种最大类间方差法与相似度相结合的改进免疫遗传方法,用于零值绝缘子红外热像分割。该方法以类间方差为适应度函数,通过自适应提取免疫疫苗、接种疫苗、免疫选择操作确保了种群的多样性,避免了普通遗传算法出现的"早熟"现象,收敛速度较快,实验验证了该方法的有效性。According to the characteristics of zero resistance insulator infrared image with poor contrast,blur target edge,large noise and the target temperature slightly higher than the environment temperature,self-adaptive median filter and mean filter are used to de-noise from the original image in this paper and enhance image by log conversion.On this basis,an improved immune genetic algorithm for zero resistance insulators infrared thermal imaging segmentation method with otsu method and similarity method combined is proposed.This method takes otsu for fitness function,ensures the population diversity by adaptive extracting immune vaccine,vaccinating vaccine,immune selection operation and avoids the traditional genetic algorithm in the＂precocious＂phenomenon,fast convergence rate,experiments verifies the validity.

关 键 词：零值绝缘子 对比度增强 类间方差法 相似度 免疫遗传算法 红外图像分割 

分 类 号：TM85[电气工程—高电压与绝缘技术]