Towards Secure Private Image Matching

Abstract

Currently, image matching is being used in many daily life applications such as content-based image retrieval (CBIR), computer vision, and near duplicate images. Hence, a number of matching methods have been developed. However, most proposed methods do not address the challenges involved when confidential images are used in image matching between two security agencies. Thus, interest to develop a secure method, particularly one that can be used in privacy-preserving image matching, is growing. This paper addresses the challenge of privacy-preserving image matching between two parties where images are confidential. The descriptor set of the queried party needs to be generated and encrypted properly with the use of a secret key at the queried party side before being transferred to the other party. We present the development and validation of a secure scheme to measure the cosine similarity between two descriptor sets. The method can work without using any image encryption, sharing, and trusted third party. We conduct several empirical analyses on real image collections to demonstrate the performance of our work.