Fast Image Search with Distrubuted Hashing

In recent years, the number of images in digital media has reached a very large size. The widespread use of image-capturing devices and Internet-based image sharing networks has a major share. In addition to the storage problem of large databases, it is another problem with these databases to be able to access them according to the content of the desired images. In recent years, the approximate nearest neighbor search method has attracted attention. This study is done to make fast and accurate search on large sized image data sets. For this purpose, studies have been investigated and a distributed LSH method has been proposed. With the distributed LSH method, instead of indexing the entire data from a single point, indexing is provided on the nodes on the cluster separately. Furthermore, instead of random functions used in the LSH method, the number of summary tables is reduced by using the basic components obtained by applying PCA method on the dataset and the classification is done in this way.