Abstract: Content-based Image Recovery (CBIR) techniques have undergone substantial research given the development of digital photographs. CBIR services are often highly expensive in terms of computational and storage costs. Therefore, choosing to outsource CBIR service to a cloud server with a lot of resources is a wise decision. However, since the cloud server cannot be completely trusted, privacy protection becomes a significant issue.
In this study, we suggest a CBIR outsourcing strategy. based utilizing a brand-new bag-of-encrypted-words (BOEW) architecture. Through color value replacement, block permutation, and intra-block pixel combination, the picture is encoded. The cloud server then uses the encrypted information to generate local histograms. picture blocks.
The cluster centres are employed as the encrypted visual words, and all of the Regional graphs grouped together. The bag-of- Pattern for coded words (BOEW) constructed in this manner A characteristics vector format or a normalized histogram of the encrypted visual words, that represents each image. The distance at Manhattan between vectors of features on a remote platform may be used to directly evaluate picture similarity. The suggested scheme's safety and precision of searches are demonstrated based on research findings and safety analysis.

Keywords: BEOW, CBIR, Digital Images, encryption

| DOI: 10.17148/IARJSET.2023.10790