Application of unsupervised image classification to semantic based image retrieval

In recent times, the ability to efficiently manage a large number of images is an important requirement of image repositories due to the increasing number of images being generated and stored in systems such as social media, digital libraries, and geographical information systems. Content Based Image Retrieval involves the management of image repositories based on the content of the images, to facilitate fast and efficient search of any desired image when needed. Although Content-Based Image Retrieval has been identified as a suitable means for supporting efficient search and retrieval of images in repositories, the presence of semantic gap in its implementation has limited its reliability, creating the move towards Semantic-Content Based Image Retrieval. This Study discusses the importance of Machine Learning in Content Based Image Retrieval, where it supports the generation of the Image representation, which is used for Indexing image repositories, and for the automatic mapping low-level image features to human language in Semantic Content Image Retrieval.