<p>This paper is an extended description of one of the basic models of descriptive image analysis that characterizes the architecture and structure of the image recognition process: a multi–level model of image analysis and recognition procedures based on the joint use of methods for combining algorithms and methods for combining fragmentary image data – partial descriptions of the object of analysis and recognition - images. The architecture, functionality, limitations and characteristics of a multilevel model for combining algorithms and initial data in image recognition are substantiated and defined. A multi–level model of image analysis and recognition procedures combines several important concepts of descriptive image analysis: multialgorithmic classifiers, dual representation of images, image models, spatial image recognition algorithms, image reduction to a form convenient for recognition. In this case, “image reduction” refers to the use of formal models and representations rather than the actual images themselves. At each level of the model, it is assumed to use a combination of selected methods/algorithms and optimization of the results. All the concepts introduced, as well as the methods of constructing descriptive algorithmic image analysis schemes for solving practical problems in accordance with the constructed process model, make up the Multialgorithmic Hierarchical Image Analysis System. This paper provides examples of the implementation of descriptive algorithmic schemes created based on the proposed early method for automated detection of pathological changes in the morphometric characteristics of the fundus using the Multialgorithmic Hierarchical Image Analysis System.</p>