Balanced principal component for 3D shape recognition using convolutional neural networks. Issue 17 (24th February 2021)

• Record Type: Journal Article
• Authors: Luo, Wenjie; Zhang, Han; Ni, Peng; Tian, Xuedong
• Is Part Of: IET image processing. Volume 14:Issue 17(2020)
• Page Start: 4468
• Page End: 4476
• Publication Date: 2021-02-24
• Subjects: principal component analysis -- shape recognition -- image classification -- image recognition -- neural nets -- learning (artificial intelligence) -- image representation -- feature extraction; computational process -- AC‐PCA module -- average compression PCA module -- regionalisation module -- BPC layer -- balanced local feature -- neural networks -- principal component analysis -- convolutional neural network -- 3D shape recognition -- balanced principal component -- state‐of‐the‐art network -- projection method -- CNN -- feature extraction task -- multidimensional balance -- one‐dimensional vector -- feature matrix; Image processing -- Periodicals; 621.36705
• DOI: 10.1049/iet-ipr.2019.0844
• Languages: English
• ISSNs: 1751-9659

Abstract

Abstract : Currently, PCA (principal component analysis) is widely used in many neural networks and has become a crucial part of the convolutional neural network (CNN) feature extraction. However, whether PCA is suitable for this process remains to be elucidated. The authors proposed a new method called balanced principal component (BPC) that generates a balanced local feature and combines with CNN as a layer to cope with the fusion problem. Specifically, BPC layer includes regionalisation module and average compression PCA (AC‐PCA) module. First, they used regionalisation module to generate some sub‐region that focuses on the local feature in each view. Secondly, the AC‐PCA module is a computational process that enlarges the feature matrix by PCA and eventually compacts the matrix to a one‐dimensional (1D) vector by AC. Next, all 1D vectors are compacted by AC to obtain a multi‐dimensional balance. Finally, they designed this layer with an end‐to‐end trainable structure to promote the feature extraction task of CNN. They addressed 3D shapes using a projection method that is pre‐trained on ImageNet and migration learning on ModelNet dataset. By comparing with the state‐of‐the‐art network, they achieved a significant gain in performance of retrieval and classification tasks.