摘要
利用多个稀疏表示分类器融合的决策信息对图像进行分类,可避免单个特征对图像分类的影响。提出一种自适应调节权重的多稀疏分类器融合图像分类方法。对原始图像分别提取3组不同特征,并训练出各自稀疏表示分类器;根据各个子分类器的准确率,通过迭代计算自适应确定各分类器最终权重;融合各子分类器的输出结果进行最终类别判断。基于Cifar-10图像数据集进行多组实验,结果表明,相对仅提取单特征的图像分类方法,该方法有效提高了图像分类准确率。
Using multiple sparse representation classifiers to combine decision information to classify images can avoid the impact of individual features on image classification. Thus, this paper proposed a multi-sparse classifier fusion image classification method with adaptive adjustment weights. Three sets of different features were extracted from the original image, and the respective sparse representation classifiers were trained. According to the accuracy of each sub-classifier, the final weight of each classifier was determined adaptively by iterative calculation. Finally, the output of each sub-classifier was merged for final category judgment. Multiple sets of experiments were performed based on the Cifar-10 image data set. The results show that the proposed can effectively improve the image classification accuracy, compared with the image classification method only extracting single feature.
引文
[1] 黄凯奇,任伟强,谭铁牛.图像物体分类与检测算法综述[J].计算机学报,2014,36(6):1-18.
[2] Wright J,Yang A Y,Sastry S S,et al.Robust Face Recognition via Sparse Representation[J].IEEE Trans Pattern Anal Mach Intell,2009,31(2):210-227.
[3] Aharon M,Elad M,Bruckstein A.K-SVD:An Algorithm for Designing Overcomplete Dictionaries for Sparse Representation[J].IEEE Transactions on Signal Processing,2006,54(11):4311-4322.
[4] Jiang Z,Lin Z,Davis L S.Label Consistent K-SVD:Learning a Discriminative Dictionary for Recognition[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2013,35(11):2651-2664.
[5] Zhang Z,Xu Y,Yang J,et al.A Survey of Sparse Representation:Algorithms and Applications[J].IEEE Access,2015,3:490-530.
[6] 顾鑫,王海涛,汪凌峰,等.基于不确定性度量的多特征融合跟踪[J].自动化学报,2011,37(5):550-559.
[7] Zhu X K,Jing X Y,Wu F,et al.Multi-Kernel Low-Rank Dictionary Pair Learning for Multiple Features Based Image Cclassification[C]// Thirty-First AAAI Conference on Artificial Intelligence,2017.
[8] Friedman A.Framing pictures:The role of knowledge in automatized encoding and memory for gist [J].Journal of Experimental Psychology General,1979,108(3):316-355.
[9] Oliva A,Torralba A.Building the gist of a scene:the role of global image features in recognition[J].Progress in Brain Research:Visual Perception,2006,155:23-36.
[10] Lowe D G.Distinctive image features from scale invariant key points[J].International Journal of Computer Vision,2004,60(2):91-110.
[11] Lindeberg T.Scale-space for discrete signals[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,1990,12(3):234-254.
[12] Dalal N,Triggs B.Histograms of Oriented Gradients for Human Detection[C]// IEEE Computer Society Conference on Computer Vision and Pattern Recognition.IEEE,2005:886-893.
[13] Patel V M,Chellappa R.Compressive Sensing[M]// Sparse Representations and Compressive Sensing for Imaging and Vision.Springer New York,2013:132-136.
[14] Corchado E.A survey of multiple classifier systems as hybrid systems[M].Elsevier Science Publishers B.V.2014.
[15] Yang M,Zhang L.Gabor Feature Based Sparse Representation for Face Recognition with Gabor Occlusion Dictionary[C]// European Conference on Computer Vision.Springer,Berlin,Heidelberg,2010.