Quantitative description of the laws of perceptual grouping by means of the principleof representational minimum description length

Abstract

This paper discusses the problem of quantitatively describing the laws of perceptive grouping developed in Gestalt psychology and their use in computer vision. For a unified description of the Gestalt laws, the principle of representational minimum description length is proposed. Psychophysical experiments have been carried out in which it is established that the probability that a person will distinguish a stimulus depends predominantly on the information content of the stimulus, expressed in the reduction of the description length (in the framework of the given representation) of the elements of the visual scene when the stimulus is described separately from the field elements. A computer model is constructed and experimentally checked that groups the structural elements of the image, based on an informational criterion and a representation that embodies some of the Gestalt laws.