A new automation process for change detection of land area is proposed. The method is based on the use of multitemporal satellite images and their histograms. The proposed discrete representation of land area in small image fragments makes it possible to detect changes automatically within each fragment. This method provides improvements of change detection accuracy and simplifies the software realization by unifying recursively performed processing. In this article the method of difference images and its adaptation for automatic applications is also analysed. In particular, a complementary pair of images is proposed as the main presentation of a difference image which allows automatic separation of the changes of ground objects without loss or distortion. The use of the histograms obtained by variations of image brightness (increasing and decreasing) provides opportunities for the assessment and experimental verification of existing approaches in the selection of automatic detection thresholds. The assessment of the implementations of the algorithms of Kittler, Kapur, Otsu, and the method of mean risk minimum demonstrate the instability of their solutions, wide variance in of results, as well as inconsistencies with the visual approaches to the selection of the binarization thresholds. A new algorithm for the automatic detection and analysis of the brightness changes of land areas is proposed. As a result, the automation of detection of each brightness change of a given land area is achieved in all available spectral channels of satellite imagery.