Efficient content-based image retrieval using deep search and rescue algorithm

Recently, content-based image retrieval (CBIR) system seems to be very challenging in research fields due to the growth of multimedia contents on the internet. Every day billions of images are uploaded over the internet. The search for a relevant image on the search engines seems quite challenging for the research community. CBIR system made this search easier as high-level image visuals which are characterized in the form of feature vectors. In this work, Deep Search and Rescue (SAR) Algorithm-based CBIR is presented for effective retrieval of relevant images. The steps involved in proposed Deep Neural Network-SAR (DNN- SAR) are pre-processing, multiple feature extraction, feature fusion, clustering and classification. Initially, Fast Average Peer Group (FAPG) filter is used to remove the noise in the pre-processing stage. Then multiple features like color, shape and texture are extracted and feature vectors are calculated. All these three features are fused into a single feature using average and weighted average techniques. Next, the fused features are clustered using adaptive Sunflower optimization (SFO) algorithm. Finally, the relevant images are retrieved using DNN-SAR optimization algorithm. The proposed work is implemented in PYTHON platform and tested on four different types of image databases, namely Corel 1 K, 1.5 K, 5 K, and Caltech-256. Thus, the simulation outcomes proved that the proposed DNN-SAR technique had improved the CBIR performance in terms of precision and recall.

Automated summary

Recently, CBIR system faces challenges due to the growth of multimedia contents on the internet. This study proposes a Deep Search and Rescue (SAR) Algorithm-based CBIR system, which effectively retrieves relevant images through pre-processing, feature extraction, fusion, clustering, and classification steps.