Efficient content based image retrieval based on Semantic Object Detection

This paper proposes an efficient content-based image retrieval scheme based on Semantic Object Detection (SOD). The feature extraction process uses quantized discrete cosine transform (DCT) blocks where the retrieval of the query image is performed using a histogram-based similarity measure. SOD aims to reduce the size of the database from which the retrieval of similar images is conducted. The use of SOD has been shown to improve the retrieval performance and outperform related work in the literature.