Due the growing popularity of image acquisition devices and INTERNET, millions of images are available. It is possible to retrieve and access public images through search engines, such as Google’s for instance. Because of the huge amount of available images in many repositories, the development of methods that can select and filter desired results is crucial. In this context classification algorithms aim to sort elements of a set using clustering and categorization. Thus it is possible to apply these algorithms to sort results from a search for images in such a manner that they are grouped according to related aspects. Hence, this work aim to create a system to filter and arrange sets of images retrieved from a search on the INTERNET. This filter intent to group similar elements in folders to aid users throughout search. The approaches in this work involve the categorization of images through dynamic clustering and descriptors that represent the information within the images.