Content based image retrieval using feature fusion technique

This paper contributes a feature fusion technique for efficient content-based image retrieval. It aims to implement an accurate feature extraction method through the proposed system model. It can be a promising candidate for image retrieval applications as it concatenates the features from deep learning architecture and local binary patterns. Experimentation is done on the benchmark Core11000 Database that has been divided into 10 classes of beaches, food, buses, flowers, dinosaurs, elephants, horses, monuments, mountains and snow, and people and villages in Africa. The proposed method achieves an average precision of 95.14% and an average recall of 9.514% while retrieving 10 images corresponding to a given query image. These results are superior to the existing methods based on deep learning models and local patterns. Also, for variations in the number of retrieved images, respective average precision, and average recall are calculated.