The paper concerns specific problems of color digital picture recognition by use of the concept of fuzzy granulation, and in addition rough information granulation. This idea employs information granules that contain pieces of knowledge about digital pictures such as location of objects as well as their size and color. Each of those attributes is described by means of linguistic values of fuzzy sets, and the shape attribute is also considered with regard to the rough sets. The picture recognition approach is focused on retrieving a picture (or pictures) from a large collection of color digital pictures (images) - based on the linguistic description of a specific object included in the picture to be recognized.