A Model for Identifying Objects in Images Based on the Concept of a Receptive Field of Neurons

The article discusses a new method for identifying objects in images, which is based on the model of registration and neural coding of radiation at the periphery of the human visual system (mainly in the retina). The proposed method combines both known data on the structure and functioning of the human visual system and modern machine learning methods for image processing and analysis. A characteristic feature of the approach is the representation of the recorded data (images) by a sample of counts formed by an ideal visualization device. The refinement of the classical statistical estimation, hypothesis testing, etc., arising from this approach and related to the issues of object recognition in images, are discussed. The synthesized numerical algorithms of the developed procedures are illustrated by the results of a computer simulation.