Evaluating images according to their content is being studied in many studies. Large number of features need to be used in order to define images according to their contents. In addition access to the content of the desired images in large scale image databases is accompanied by many problems that must be overcome. This study was carried out in order to access the relevant images on the large volume image databases quickly and correctly. Features of images are stored in a table form in relational databases and database queries are made according to the index structure created by using Local Sensitive Hashing method and similar images are tried to be found quickly.