Publication Type

Journal Article

Publication Date



Bag-of-features (BoF) is one of the most well-known methods used to represent digital image features because of its simplicity and efficiency. A variety of improved algorithms have been employed to enhance the performance of BoF in characterization. However, challenges in the application of BoF in the field still exist. This study focused on BoF by decomposing local features and presented a novel framework for BoF on the basis of low-rank and sparse matrix decomposition to obtain a more robust and discriminative digital image classification. First, the local feature matrix of a digital image is decomposed into a low-rank matrix and a sparse matrix. Then, the BoF model was constructed in each part. Finally, the multiple kernel learning method was applied to combine the two models and the digital images were classified by using the support vector machine. Compared with existing methods in five public data sets, results show that the method proposed in this study is superior to the baseline algorithm and other coding algorithms by improving local features, with an improved classification performance of 17.68% in maximum and 0.01% in minimum. Compared with similar methods (such as leveraging the low-rank and sparse matrix decomposition and group sparse coding for image classification), this method is superior, with an improved classification performance of 2.76% in maximum and 0.08% in minimum, and obtains the highest average correct rate of classification. Therefore, the proposed method in this study is effective in improving the BoF in the feature extraction stage and has a better image classification performance.


Bag-of-features, Digital image classification, Low-rank and sparse matrix decomposition, Multiple kernel learning


Databases and Information Systems | Theory and Algorithms


Journal of Digital Information Management





First Page


Last Page





Digital Information Research Foundation

Creative Commons License

Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.

Additional URL