基于GA-FKCN聚类的图像分割-应用数学专业论文.docxVIP

鞭南交通丈学磺士研究生学伎论文 鞭南交通丈学磺士研究生学伎论文 繁l页 摘 要 聚类分析是按照～定要求和蕊律进行事物分类的一种数学方法。 客观事物阀的界限往往不一定很清晰，因此将模糊理论遴用到聚类分 析中会更囱然合理。囊Zadeh于1965年提出模糊集理论聪，人们就搿 楚蔫攘凝貔方法寒楚建莱荣毒模糊经懿分类翊蘧，并褥爨了广泛豹发 展，其运粥领域涉及剃医学诊断、模式识别和图像处理替。 本文讨论了模糊槊类和图像分割技术的料究现状，介绍了遗传算 法的发鼹历史和基本概念、特点、基本原理和流程，分析了模糊 Kohonen浆类圈络(模穰基壅缀淤袈)算法。在踅基懿上，本文箍懑 ～种改进的模期图像分割算法，包括预处理和分割过程。预处理阶段 给出了新的隶属函数嗣改进的模糊增强函数，并与传统Pal算法对比， 分橱了鼗媳数的改避链。分割道程提出了蕊于GA—FKCN(Genetic Algorithm—Fuzzy Kohonen Clustering Network斡聚焱冀法，结合 了遗传算法的全局性和模糊Kohonen聚类网络的自组织性。最后通过 图像分割实验证明本文算法的效果优于传统的模糊图像分割。 美键谲： 模襁聚类；遗傣算法；棒经隧络；鹫豫分裁： 西南交通大学硕士研究生学位论文 西南交通大学硕士研究生学位论文 第1 I页 Abst ract Clustering i s a mathematic method which classifies things accordlng to some criteria and rules．Since the borderline of the entities i s often indefinite，it i s more natural to utilize ℃fuzzy theory to clustering．After Zadeh putting forward Fuzzy Theory，it i s popufar that using fuzzy theory to deal with the classifi cation with fuzzy characters，then widely developing， and applying to the fields including medi cal diagnoses，pattern recognition and image processing etc．． In thi s di ssertat ion，the research status of fuzzy clustering and image segmentation is studied：a brief review of the hi story，basi c theorY and flow of genetic algorithm i s presented； fuzzy Kohonen e 1usteri ng network theory i S i11ustrated．Based on these theories above，an improvement algorithm of image segmentation including pretreatment and the course of segmentation i s presented． In the pretreatment， a new membershiP function and an improved fuzzy enhancement function are given，and then the ametioration of the improved algorithm i s analYzed aCCording to the comparison with traditj ohal Pal method．During the tourse of image segmentation，a clustering algorithm based on GA——FKCN(Genetic A190rithm-FuzzY Kohonen Cl ustering Network)，which combine the character of ensemble with the self-organization of fuzzY Kohonen clustering network． FihallY．the image segmenta