Combining Extreme Learning Machine, RF and HOG for Feature Extraction

被引:2
作者
Ouyang, Jianquan [1 ]
Hu, Qianlei [1 ]
机构
[1] Xiangtan Univ, Key Lab Intelligent Comp & Informat Proc, Xiangtan 411105, Hunan, Peoples R China
来源
2017 IEEE THIRD INTERNATIONAL CONFERENCE ON MULTIMEDIA BIG DATA (BIGMM 2017) | 2017年
关键词
neural network; Histogram of Oriented Gradient; random forest; feature extraction; Extreme Learning Machine;
D O I
10.1109/BigMM.2017.60
中图分类号
TP39 [计算机的应用];
学科分类号
081203 ; 0835 ;
摘要
Feature extraction and classifier is crucial for content-based image retrieve and analysis. In this paper, a novel method for handwritten numeral image extraction is proposed based on Random Forest(RF) and Histogram of Oriented Gradient(HOG). The main contribution of the proposed method is to consider the advantage of HOG and RF. Further, our method extract the impactful information of image, and reduce the input dimension. Therefore, the method can effectively improve the accuracy of handwriting recognition. And the training speed of classifier is faster than multi-layer neural network. We performed the proposed method on MNIST and USPS datasets. The experimental results show that the proposed method has higher accuracy and faster training speed than HOG-ELM and ML-ELM.
引用
收藏
页码:419 / 422
页数:4
相关论文
共 13 条
[1]  
AKHAND MAH, 2016, COMPUTER AND COMMUNI, P311, DOI DOI 10.1109/ICCCE.2016.73
[2]   Histograms of oriented gradients for human detection [J].
Dalal, N ;
Triggs, B .
2005 IEEE COMPUTER SOCIETY CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION, VOL 1, PROCEEDINGS, 2005, :886-893
[3]  
de Chazal P, 2015, INT CONF ACOUST SPEE, P2165, DOI 10.1109/ICASSP.2015.7178354
[4]  
Guo Peng, 2016, RES DEEP CONVOLUTION
[5]  
Huang GB, 2004, IEEE IJCNN, P985
[6]  
Kasun LLC, 2013, IEEE INTELL SYST, V28, P31
[7]  
Keysers D., 2016, IEEE T PATTERN ANAL
[8]  
Kim Gab Soon, 2015, [Journal of IKEEE, 전기전자학회논문지], V19, P526, DOI 10.7471/ikeee.2015.19.4.526
[9]  
Li Tianqiang, 2008, HANDWRITTEN DIGIT RE
[10]  
SETHY A, 2016, COMPUTING COMMUNICAT, P1099