基于粒子群的支持向量机图像识别  

Support vector machine image recognition based on particle swarm

作　　者：韩晓艳[1,2] 赵东[3] 

机构地区：[1]吉林农业大学信息技术学院,吉林长春130118 [2]长春理工大学光电信息学院信息工程分院,吉林长春130012 [3]长春师范大学计算机科学与技术学院,吉林长春130032

出　　处：《液晶与显示》2017年第1期69-75,共7页Chinese Journal of Liquid Crystals and Displays

基　　金：国家自然科学基金(No.61101155);吉林省自然科学基金(No.20140101184JC);长春市科技计划资助项目(No.2012091);吉发改高技(No.2014817);吉林省教育厅(No.2016392)~~

摘　　要：为了实现对田间水稻缺素的精准识别,构建一个图像识别系统。对该系统所采用的图像采集、图像分割、基于支持向量机图像分类等算法进行研究。首先,根据田间水稻的缺素现象进行图像采集和处理。然后提取图像与氮元素相关的颜色特征。在分析比较SVM算法对图像分割的基础上,提出一种基于改进粒子群算法进行SVM参数优化算法模型(即IPSO-SVM)。最后,对实验进行设置,对算法模型与其他算法进行测试对比。实验结果表明:对水稻缺素诊断的准确率达到95.45%,基本满足田间水稻缺素的科学诊断要求。In order to realize the precise identification of the missing elements in the field of rice,an image recognition system was constructed.The image acquisition,image segmentation and image classification based on support vector machine are studied in this system.Firstly,image acquisition and processing were carried out according to the phenomenon of deficiency of rice in the field.Then the color features were extracted related to the image of nitrogen elements.On the basis of analyzing and comparing the SVM algorithm to image segmentation,an improved particle swarm optimization algorithm is proposed to optimize the SVM parameters（i.e.IPSO-SVM）.At last,the experiment is set up,and the algorithm model is tested and compared with other algorithms.The experimental results show that the accuracy rate of the diagnosis of rice deficiency is 95.45%,which basically meets the requirements of the scientific diagnosis of the deficiency of rice in the field.

关 键 词：图像分割 支持向量机 粒子群 缺素 

分 类 号：TP394.1[自动化与计算机技术—计算机应用技术] TH691.9[自动化与计算机技术—计算机科学与技术]