PROJECT TITLE :

BOEW: A Content-Based Image Retrieval Scheme Using Bag-of-Encrypted-Words in Cloud Computing

ABSTRACT:

Because of the explosive growth of digital images, techniques known as content-based image retrieval, or CBIR, have been the subject of a lot of research. The CBIR service, in general, requires a significant investment of both computational and storage resources. As a result, contracting the CBIR service out to a cloud server that has access to a vast amount of resources is a sensible option to consider. On the other hand, given that it is impossible to put complete faith in the cloud server, protecting users' privacy becomes a significant challenge. An outsourced CBIR scheme that is based on a novel bag-of-encrypted-words (BOEW) model is what we propose in this piece of research. Substitution of color values, block permutation, and intra-block pixel permutation have been used to encrypt the image. The cloud server will then proceed to derive the local histograms by using the encrypted image blocks as input. Each of the local histograms is grouped together into a cluster, and the centers of the clusters are what are used as the encrypted visual words. The bag-of-encrypted-words (BOEW) model is constructed in this manner in order to represent each image by a feature vector, which is equivalent to a normalized histogram of the encrypted visual words. On the cloud server side, the Manhattan distance between feature vectors can be used to directly measure the degree to which two images are similar to one another. The experimental findings and security analysis carried out on the proposed scheme provide evidence of the accuracy and safety of the search.