Abstract
Canonical Correlation Analysis (CCA) is a classic multivariate statistical technique, which can be used to find a projection pair that maximally captures the correlation between two sets of random variables. The present paper introduces a CCA-based approach for image retrieval. It capitalizes on feature maps induced by two images under comparison through a pre-trained Convolutional Neural Network (CNN) and leverages basis vectors identified through CCA, together with an element-wise selection method based on a Chernoff-information-related criterion, to produce compact transformed image features; a binary hypothesis test regarding the joint distribution of transformed feature pair is then employed to measure the similarity between two images. The proposed approach is benchmarked against two alternative statistical methods, Linear Discriminant Analysis (LDA) and Principal Component Analysis with whitening (PCAw). Our CCA-based approach is shown to achieve highly competitive retrieval performances on standard datasets, which include, among others, Oxford5k and Paris6k.
Highlights
The past two decades have witnessed an explosive growth of online image databases.This growth paves the way for the development of visual-data-driven applications, but at the same time poses a major challenge to the Content-Based Image Retrieval (CBIR) technology [1].Traditional approaches to CBIR mostly rely on the exploitation of handrafted scale- and orientation-invariant image features [2,3,4,5,6], which have achieved varying degrees of success
Convolutional Neural Network (CNN) models are usually trained for purposes different from CBIR, it is known [9] that features extracted from modern deep CNNs, commonly referred to as Deep Learning (DL) features, have great potential in this respect as well
Motivated by the consideration of computational efficiency and affordability as well as the weaknesses inherent in the existing preprocessing methods, we develop and present in this paper a new image retrieval method based on OTS deep CNNs
Summary
The past two decades have witnessed an explosive growth of online image databases.This growth paves the way for the development of visual-data-driven applications, but at the same time poses a major challenge to the Content-Based Image Retrieval (CBIR) technology [1].Traditional approaches to CBIR mostly rely on the exploitation of handrafted scale- and orientation-invariant image features [2,3,4,5,6], which have achieved varying degrees of success. The past two decades have witnessed an explosive growth of online image databases This growth paves the way for the development of visual-data-driven applications, but at the same time poses a major challenge to the Content-Based Image Retrieval (CBIR) technology [1]. A main advantage of such methods is the low implementation cost [15,16], which is largely attributed to the direct adoption of pre-trained CNNs. Performance-wise, they are comparable to the state-of-the-art traditional methods that rely on handcrafted features. A top representative from this category is the end-to-end learning framework proposed in [20] It outperforms most existing traditional and OTS-CNN-based methods on standard testing datasets; this performance improvement comes at the cost of training a complex triple-branched CNN using a large dataset, which might not always be affordable in practice
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
