Bioplausible Multiscale Filtering in Retinal to Cortical Processing as a Model of Computer Vision

Abstract

Visual illusions emerge as an attractive field of research with the discovery over the last century of a variety of deep and mysterious mechanisms of visual information processing in the human visual system. Among many classes of visual illusion relating to shape, brightness, colour and motion, "geometrical illusions" are essentially based on the misperception of orientation, size, and position. The main focus of this paper is on illusions of orientation, sometimes referred to as "tilt illusions", where parallel lines appear not to be parallel, a straight line is perceived as a curved line, or angles where lines intersect appear larger or smaller. Although some low level and high level explanations have been proposed for geometrical tilt illusions, a systematic explanation based on model predictions of both illusion magnitude and local tilt direction is still an open issue. Here a neurophysiological model is expounded based on Difference of Gaussians implementing a classical receptive field model of retinal processing that predicts tilt illusion effects.

Bioplausible multiscale filtering in retino-cortical processing as a mechanism in perceptual grouping

“…The patterns investigated have some similarities to Brightness/Lightness Illusions such as Irradiation [45,62], Simultaneous Brightness Contrast (SBC) [63][64][65] and White's effect [15,63,[65][66][67][68][69], but the difference between the explanation of these two subclasses of illusion is that for Tilt Illusions we seek for the prediction of tilt not the changes of Brightness/Lightness profile used to describe Brightness induction effects. The patterns under investigation seem to have mid-to high-level perceptual explanations, but in our previous investigations [14,[48][49][50] we have shown that the illusory tilt cues in these patterns are caused by multiscale retinal/cortical encoding by the simple cells and the Lateral Inhibition among them. We demonstrated how a low-level explanation of these patterns at multiple scales reveals the tilt cues in the local processing of the pattern, followed by higher levels of processing in the retina and the cortex for the integration of the local tilt cues for the final percept.…”

“…The patterns explored in this paper are 'second-order Tilt' Illusions [14] (Tile Illusions) involving the enhancement of contrast between textural elements of a background such as a checkerboard, for example Café Wall and Bulging checkerboard illusions. In the Café Wall illusion, the illusory tilt percept is the result of mortar lines between shifted rows of black and white tiles.…”

“…In this study, we further explore the neurophysiological model of multiple-scale low-level filtering developed by Nematzadeh et al [14], based on the circular centre and surround mechanism of classical receptive field (CRF) in the retina. For filtering, a set of the Differences of Gaussians (DoGs) at multiple scales is used to model the multiscale retinal ganglion cell (RGC) responses to the stimulus.…”

The Café Wall Illusion: Local and Global Perception from Multiple Scales to Multiscale

“…A low-level filtering approach [17,19,22,23] have been explored here modelling the retinal/cortical simple cells in our early vision for revealing the tilt cues involved in the local and global perception of the Café Wall stimulus. The model has an embedded processing pipeline utilizing Hough transform to quantify the degrees of the inducing tilts appeared in the low-level representation for the stimulus in our model, referred to as the edge map at multiple scales.…”

“…The model here [17][18][19][20][21][22] is a most primitive implementation for the contrast sensitivity of RGCs based on classical circular center and surround organization of the retinal RFs [33,34]. The output of the model is a simulated result for the responses of the retinal/cortical simple cells to a stimuli/image.…”