ANALYSIS OF A MODEL OF VISION SYSTEM CELLS ON THE BASE OF THE MINIMUM REPRESENTATION LENGTH PRINCIPLE FOR IMAGE DESCRIPTION
ITMO University, Saint Petersburg, 197101, Russian Federation; Professor
I. P. Gurov
ITMO University, 197101, Saint Petersburg, Russian Federation; Department Head
A. . Averkin
ITMO University; student
Abstract. Theoretical analysis of an informational model of visual cortex cell selective to spatial patterns is carried out. Methods of textural segmentation based on accepted model are demonstrated to diminish representation length of image description. Explicit application of image description length as an objective function is shown to allow for the model improvement.
Keywords: visual cortex, model of neuron, minimal description length, textural segmentation.