Can multimedia databases handle the growing demand for efficient content-based retrieval? This survey examines the challenges and solutions for similarity searches in high-dimensional spaces, a key component of multimedia database systems. The paper reviews various index structures and algorithms designed to enhance query processing in these complex environments. As multimedia data becomes increasingly prevalent in fields like medicine, CAD, and geography, content-based retrieval has emerged as a critical functionality. The survey explores how feature transformations are used to convert multimedia object properties into high-dimensional feature vectors, effectively transforming similarity searches into proximity searches in the feature space. It highlights the significant performance improvements achieved by new index structures and algorithms. This overview provides valuable insights into the state-of-the-art techniques for querying multimedia databases. By addressing the inherent problems of processing queries in high-dimensional spaces, the described approaches pave the way for more efficient and scalable multimedia information retrieval systems, enabling users to quickly and accurately find similar objects within large databases. These advancements have significant implications for various multimedia applications and data-driven research.
Published in ACM Computing Surveys, this work fits perfectly with the journal's scope of providing comprehensive overviews of important topics in computer science. By surveying the current state of similarity search techniques in high-dimensional spaces, this paper serves as a valuable resource for researchers and practitioners in the field, aligning with the journal’s goal of promoting knowledge dissemination and advancement in computing.