Th e article discusses the possibility to use singular values to classify areal objects according to their appearance on the Landsat-7 digital images made in panchromatic, green, red, and infrared light ranges. It is proposed to use the parameters of a0 and a1 straight lines that approximate singular values of gray level matrices as a characteristic that allows classifying the images of areal objects. Methods of object identification has been worked out by using a test object that resulted in verification of control objects. For this purpose, there were used the images of water surface of Irtysh river, built-up territories of the city of Khanty-Mansiysk, as well as forests and marshes. It is shown that a field of characteristics of the objects considered represents nonoverlapping enclosed areas, and does not depend on changes of general background of the images.