A Novel Replication-Less Image Retrieval Method from Cloud Platforms using Divergence Features

Abstract

Real-time multimedia applications provide visual and audible services collaborated with the cloud platform for heterogeneous user requirements. Multi-source information storage and fusion satisfy the user demands through the applications. Contrarily, replication is a common issue demanding high space and time, increasing computation and retrieval time. For addressing the retrieval issues in replicated image storage, this article introduces a Divergent Feature-induced Extension (DFIE) method for large cloud platforms. The proposed method identifies the divergences alone in the input features correlated with the stored ones, in different feature extracted instances. In the feature divergent analysis, the deep recurrent learning paradigm is utilized. The iterations are used for identifying non-correlating features and their deviation for verifying non-replicable images. The process pursues region-based segmentation for a feature and edge-based divergence and similarity identification. The proposed method’s performance is analyzed using the metrics detection accuracy, complexity, and computing time.