• 中国计算机学会会刊
  • 中国科技核心期刊
  • 中文核心期刊

Computer Engineering & Science ›› 2021, Vol. 43 ›› Issue (04): 729-737.

Previous Articles     Next Articles

Saliency detection based on multi-feature fusion convolutional neural network

ZHAO Ying-ding1,YUE Xing-yu2,YANG Wen-ji1,4,ZHANG Ji-hao3,YANG Hong-yun1,4#br#

#br#
  

  1. (1.School of Software,Jiangxi Agricultural University,Nanchang 330045;

    2.School of Computer and Information Engineering,Jiangxi Agricultural University,Nanchang 330045;

    3.School of Foreign Languages,Huazhong University of Science and Technology,Wuhan 430074;

    4.Key Laboratory of Agricultural Information Technology of 
    Colleges and Universities in Jiangxi Province,Nanchang 330045,China)
  • Received:2019-12-26 Revised:2020-06-02 Accepted:2021-04-25 Online:2021-04-25 Published:2021-04-21
  • Supported by:
    国家自然科学基金(61462038,61562039);江西省教育厅科技计划项目(GJJ190217)

Abstract: With the development of deep learning technology and the prominent performance of con- volutional neural networks in many computer vision tasks, deep saliency detection methods based on convolutional neural networks have become the mainstream methods in saliency detection. However, the convolutional neural network is limited by the size of the convolution kernel, which can only extract features in a small region at the bottom of the network, and cannot detect the objects that are not notable in the region but are globally remarkable. On the other hand,the convolutional neural network can obtain the global information of the image by stacking the convolutional layers, but when the information is transferred from shallow layers to deep layers, it will lead to the loss of information, and stacking too deep will also make the network difficult to optimize. For these reasons, a saliency detection method based on multi-feature fusion convolutional neural network is proposed. In this method, the convolutional neural network is enhanced by several local feature enhancement modules and global context mo- deling modules. Specifically, the local feature enhancement module is used to increase the feature extraction range, and the global information of the feature map is obtained by global context modeling, which effectively suppresses the interference of objects in the region which are notable in the region but not significant in the whole image to the saliency detection. It can also extract multi-scale local features and global features simultaneously for salient detection, which effectively improves the accuracy of detection results. Finally, through experiments, the effectiveness of the proposed method is verified and compared with other 11 saliency detection methods. The results show that the proposed method can improve the accuracy of saliency detection and outperform the other 11 methods involved in the comparison.


Key words: saliency detection, multi-scale, convolutional neural network, local feature enhancement, global feature modeling