This paper presents the MEDIM software system created for the study of a number of methods used in the content–based visual retrieval, based on color characteristics. The problems studied with the help of this system are content–based image query and extraction of color regions. The study of the implemented algorithms was conducted from the point of view of the retrieval quality and complexity, in different cases of transformation and quantization of color spaces, on a database with color medical images from the digestive apparatus sphere, collected with the help of an endoscope.