|本期目录/Table of Contents|

[1]曹冬寅,王 琼*,张兴敢.基于稀疏重构残差和随机森林的集成分类算法[J].南京大学学报(自然科学),2016,52(6):1127.[doi:10.13232/j.cnki.jnju.2016.06.017]
 Cao Dongyin,Wang Qiong*,Zhang Xinggan.Ensemble classification method based on sparse reconstruction residuals and random forest[J].Journal of Nanjing University(Natural Sciences),2016,52(6):1127.[doi:10.13232/j.cnki.jnju.2016.06.017]
点击复制

基于稀疏重构残差和随机森林的集成分类算法
()
     

《南京大学学报(自然科学)》[ISSN:0469-5097/CN:32-1169/N]

卷:
52
期数:
2016年第6期
页码:
1127
栏目:
出版日期:
2016-12-01

文章信息/Info

Title:
Ensemble classification method based on sparse reconstruction residuals and random forest
作者:
 曹冬寅王 琼*张兴敢
南京大学电子科学与工程学院,南京,210023
Author(s):
 Cao DongyinWang Qiong*Zhang Xinggan
School of Electronic Science and Engineering,Nanjing University,Nanjing,210023,China
关键词:
 稀疏表示图像分类算法重构算法随机森林
Keywords:
 sparse representationimage classification algorithmreconstruction algorithmrandom forest
分类号:
TP181
DOI:
10.13232/j.cnki.jnju.2016.06.017
文献标志码:
A
摘要:
传统的基于稀疏表示的图像分类算法,通常根据稀疏重构后类残差向量的l2范数得到分类判决.在复杂情况下,各类残差向量l2的范数差别可能并不明显,从而导致分类器作出错误判决.提出了一种基于稀疏表示和随机森林的集成分类方法,通过稀疏表达字典对图像进行重构,提取各类残差图像的l2范数组成特征向量,并引入随机森林进行分类判决,有效地提升了算法基于类残差向量的判决能力.在手写数字数据库MNIST上的实验结果表明,在训练样本数较少的情况下,提出的基于稀疏表示和随机森林的集成学习分类方法与目前主流的SVM分类方法及随机森林方法进行比较,识别率有较为明显的提高,具有良好的鲁棒性.
Abstract:
Based on the sparse representation computed by l2­minimization and ensemble learning,we propose a general classification algorithm for image classification.This new framework provides new insights into two crucial issues in image classification:feature extraction and classification accuracy.Since it was proposed,random forest has become a well­known data analysis method,and it has been applied to a wide variety of scientific areas.As the random forest classification has a good performance and high stability on classification,in this paper,we choose random forest as an ensemble learning classifier.The classifier based on sparse representation classified the test sample by calculate its l2 norm of residual vector between its real values and its reconstructed values.While in some cases,due to the difference of the residuals are very small,it is hard to decide the right class that the test sample belongs.We have proposed a reconstruction algorithm of sparse representation to extract image features and classify the images by random forest classifier.First,a learning dictionary is obtained based on the trained image data set.We generate a sparse vector on the over­complete dictionary,and then calculate the residuals between the real values and the reconstructed values of the training samples.The residual vector is used as the training sample of the random forest classifier.Finally the image is classified by the trained random forest classifier.Random forests are respectively constructed based on residuals,and the classification result is decided by voting strategy.Our Experiments use the standard digital database MNIST as the image recognition database.The recognition rate of the method proposed in this paper is obviously prior to some other popular classification methods,such as SVM.We use MATLAB to finish the research experiment.The experimental results indicate that the method we proposed has better performance than methods based on random forest and sparse representation respectively.Besides,this method has the stability of the result of the classification and good noise robustness.

参考文献/References:

[1] Wright J,Yang A Y,Ganesh A,et al.Robust face recognition via sparse representation.IEEE Transactions on Pattern Analysis and Machine Intelligence,2009,31(2):210-227.
[2]  陈 波,詹永照,成科扬.基于字典优化的稀疏表示的视频镜头分类.计算机应用研究,2012,29(6):2375-2378.(Chen B,Zhan Y Z,Cheng K Y.Video shot classification based on sparse representation of dictionary optimized.Application Research of Computers,2012,29(6):2375-2378.)
[3]  宋相法.基于稀疏表示和集成学习的若干分类问题研究.博士学位论文.西安:西安电子科技大学,2013.(Song X F.Study of classification problems based on sparse representation and ensemble learning.Ph.D.Dissertation.Xi’an:Xidian University,2013.)
[4]  Gan B,Zheng C H,Zhang J,et al.Sparse representation for tumor classification based on feature extraction using latent low­rank representation.BioMed Research International,2014,2014,543-546.
[5]  徐文强,高以成,周煜坤.基于稀疏表示的多尺度目标跟踪算法.计算机应用,2013,33(A02):179-182.(Xu W Q,Gao Y C,Zhou Y K.Multi­scale object tracking algorithm based on sparse representation.Journal of Computer Applications,2013,33(A02):179-182.)
[6]  Zawbaa H M,Hazman M,Abbass M,et al.Automatic fruit classification using random forest algorithm.In:Hybrid Intelligent Systems (HIS).The 11th International Conference of the USA Fuzzy Information Processing Society.California:IEEE Press,2014,164-168.
[7]  张春霞,张讲社.选择性集成学习算法综述.计算机学报,2011,8:1399-1410.(Zhang C X,Zhang J S.A Survey of selective ensemble learning algorithms.Chinese Journal of Computers,2011,8:1399-1410.)
[8]  王丽婷,丁晓青,方 驰.基于随机森林的人脸关键点精确定位方法.清华大学学报(自然科学版),2009,4:543-546.(Wang L T,Ding X Q,Fang S.Accurate localization of facial feature points based on random forest classifier.Tsinghua University(Sciences and Technology),2009,4:543-546.)
[9]  Amini S,Homayouni S,Safari A.Semi­supervised classification of hyperspectral image using random forest algorithm.In:Geoscience and Remote Sensing Symposium (IGARSS).The 14th International Conference of the Canada Fuzzy Information Processing Society.Quebec City:IEEE Press,2014,2866-2869.
[10]  Wang L,Wang Y,Zhao D.Building emerging pattern (EP) random forest for recognition.In:Image Processing (ICIP).The 14th International Conference of the Europe Image Processing  Society.Belgium City:IEEE Press,2010,1457-1460.
[11]  宋相法,焦李成.基于稀疏表示及光谱信息的高光谱遥感图像分类.电子与信息学报,2012,34(2):268-272.(Xu X F,Jiao L C.Classification of hyperspectral remote sensing image based on sparse representation and spectral information.Journal of Electronics & Information Technology,2012,34(2):268-272.)
[12]  Palmer D S,O’Boyle N M,Glen R,et al.Random forest models to predict aqueous solubility.Journal of Chemical Information and Modeling,2007,471:362-337.
[13]  张立朝,毕笃彦,查宇飞等.基于二值随机森林的目标跟踪算法.计算机应用研究,2014,31(5):1571-1573.(Zhang L Z,Bi D Y,Zha Y F,et al.Target tracking based on binary random forest.Application Research of Computers,2014,31(5):1571-1573.)
[14]  Piir G,Sild S,Maran U.Classifying bio­concentration factor with random forest algorithm,influence of the bio­accumulative vs.non­bio­accumulative compound ratio to modeling result,and applicability domain for random forest model.SAR and QSAR in Environmental Research,2014,2512.
[15]  崔建明,刘建明,廖周宇.基于SVM算法的文本分类技术研究.计算机仿真,2013,2:299-302+368.(Cui J M,Liu J M,Liao Z Y.Research of Text categorization based on support vector machine.Computer Simulation,2013,2:299-302+368.)
[16]  李 响,谭南林,李国正等.基于Zernike矩的人眼定位与状态识别.电子测量与仪器学报,2015,3:390-398.(Li X,Tan N L,Li G Z,et al.Eye location and status recognition based on Zernike moments.Journal of Electronic Measurement and Instrumentation,2015,3:390-398.)

相似文献/References:

备注/Memo

备注/Memo:
基金项目:毫米波国家重点实验室开放课题(K201514)
收稿日期:2016-03-16
*通讯联系人,E­mail:wangqiong@nju.edu.cn
更新日期/Last Update: 2016-11-21