Visual Illusion Induced by Uncertain Environment in Neural Dynamics Cognitive System

摘要

The visual signal processing mechanism of humans and mammals is precise and complex, better than most current machine vision systems. Visual illusion can show the cognitive functions and mechanisms of the visual system from a unique perspective which are of great significance to the development of computer vision. Firstly, the Gabor filter perfectly simulates the direction and spatial frequency selectivity of the visual cortex cells, thus successfully explaining why the Einstein-Monroe picture has two interpretations when viewing distance is changed and verifying the multi-scale characteristics of the visual neural network. Secondly, by means of nonlinear dynamics, the visual neural dynamics system based on the multi-scale background environment is established to reproduce optical illusions like Herman grid illusion. By understanding the mechanism of visual illusion, it can provide convenient experimental means for realizing the multi-modal information integration of multi-scale neural network and theoretical research of memory fusion model.