RÉSUMÉ

The problem on creating algorithms for linear clusterization of points in multidimensional space is considered. The approach is based on the results of experiments performed by the authors in the field of projective and pseudoinverse matrix perturbation and on recurrent estimation of point sets accumulation. This method of clusterizing the points is applied to problems on image segmentation. The work contains the results of its application to tasks on processing data for populated areas photos made from space. The brief literature review of image fragmentation by clusterizing techniques as well as classification of existing methods have been presented. The place and category of the proposed method have been identified.