Richer feature for image classification with super and sub kernels based on deep convolutional neural network. (August 2017)

• Record Type: Journal Article
• Authors: Tang, Pengjie; Wang, Hanli
• Is Part Of: Computers & electrical engineering. Volume 62(2017)
• Page Start: 499
• Page End: 510
• Publication Date: 2017-08
• Subjects: Deep convolutional neural network -- Super convolutional kernel -- Sub convolutional kernel -- Parallel crossing -- Image classification; Computer engineering -- Periodicals; Electrical engineering -- Periodicals; Electrical engineering -- Data processing -- Periodicals; Ordinateurs -- Conception et construction -- Périodiques; Électrotechnique -- Périodiques; Électrotechnique -- Informatique -- Périodiques; Computer engineering; Electrical engineering; Electrical engineering -- Data processing; Periodicals; Electronic journals; 621.302854
• DOI: 10.1016/j.compeleceng.2017.01.011
• Languages: English
• ISSNs: 0045-7906

Abstract

Highlights: The concepts of super convolutional kernel and sub convolutional kernel are proposed inspired by the process that human observes objects. A novel Parallel Crossing Deep Convolutional Neural Network (PC–DCNN) model is designed to simulate the human binocular mechanism for extracting richer information. A framework named multi-scale PC–DCNN is deployed based on super convolutional kernel and sub convolutional kernel for image classification. Graphical abstract: Abstract: Deep convolutional neural network (DCNN) has obtained great successes for image classification. However, the principle of human visual system (HVS) is not fully investigated and incorporated into the current popular DCNN models. In this work, a novel DCNN model named parallel crossing DCNN (PC–DCNN) is designed to simulate HVS with the concepts of super convolutional kernel and sub convolutional kernel being introduced. Moreover, a multi-scale PC–DCNN (MS-PC-DCNN) framework is designed, with which a batch of PC–DCNN models are deployed and the scores from each PC–DCNN model are fused by weighted average for the final prediction. The experimental results on four public datasets verify the superiority of the proposed model as compared to a number of state-of-the-art models.