4.6 Article

Discriminative convolution sparse coding for robust image classification

期刊

MULTIMEDIA TOOLS AND APPLICATIONS
卷 81, 期 28, 页码 40849-40870

出版社

SPRINGER
DOI: 10.1007/s11042-022-12395-0

关键词

Convolutional sparse coding; Sparse representation; Classification; Dictionary learning

向作者/读者索取更多资源

Convolutional Sparse Coding (CSC) is a popular model in signal and image processing, and this paper proposes a novel discriminative model based on CSC for image classification. Experimental results demonstrate the superior performance of the proposed method.
Convolutional Sparse Coding (CSC) is a popular model in the signal and image processing communities, resolving some limitations of the traditional patch-based sparse representations. However, most existing CSC algorithms are suited for image restoration. Also, in some CSC-based classification methods, the CSC model is only used as a feature extractor and so other classifiers are needed for classification. In this paper, we present a novel discriminative model based on CSC for image classification. The proposed method, discriminative local block coordinate descent (D-LoBCoD), is based on extending the LoBCoD algorithm by incorporating the classification error into the objective function that considers the performance of a linear classifier and the representational power of the filters, simultaneously. Thus, in the training phase, in each iteration, after updating the sparse coefficients and convolutional filters, we minimize the classification error by updating the parameters of the classifier according to the class label information of the training samples. Also, in the test phase, the label of the query image is determined by the trained classifier. To demonstrate the performance of the proposed method, we conduct extensive experiments on image data sets in comparison with state-of-the-art classification methods. The experimental results show that our method outperforms other competing methods in most cases. Further, we demonstrate that our proposed method is less dependent on the number of training samples because of capturing more representative information from the corresponding images. Thus our proposed method can work better than other methods on all small databases that have fewer samples.

作者

我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。

评论

主要评分

4.6
评分不足

次要评分

新颖性
-
重要性
-
科学严谨性
-
评价这篇论文

推荐

暂无数据
暂无数据