Content based Image retrieval using FAST Machine learning approach in Cloud computing

Content-based image retrieval (CBIR) is a retrieval process that focuses on the visual content of images rather than relying on text metadata. In CBIR, the search is performed based on the inherent visual characteristics of the images themselves, enabling more precise and contextually relevant image retrieval compared to traditional text-based approaches. The medical, multimedia, and surveillance industries have all expressed interest in CBIR. This review research covers all of the different CBIR strategies, including feature extraction, feature representation, similarity measurement, and relevance feedback. This paper presents several feature extraction techniques, including colour-based, texture-based, and shape-based ones. Additionally covered are methods based on deep learning that use histograms, bags of visual words, and other techniques for feature representation. Other metrics for determining similarity, such as Euclidean distance, Cosine similarity, and Jaccard similarity, are also included in the paper. Also presented are pertinent feedback techniques like query expansion and refinement. The paper provides information on current trends and potential new directions for CBIR research, as well as the advantages and disadvantages of each technique. According to the evaluation, CBIR has the potential to grow into a powerful tool for image retrieval across a variety of fields. However, problems like the semantic gap and scalability still exist. A full overview of CBIR techniques and their applications is provided in this review paper’s conclusion. The paper’s objectives are to assist practitioners and researchers in understanding the current state of CBIR research and to provide suggestions for further investigation. With technical advancements, CBIR has the potential to overcome present challenges and improve upon its performance, making it a crucial tool for image retrieval across numerous industries.