Modelling contrast sensitivity as a function of retinal illuminance and grating area

Area of Science:

• Vision Science
• Human Physiology
• Sensory Neuroscience

Background:

• Human contrast sensitivity is well-modeled at high light levels.
• Extending these models to low light requires accounting for quantal noise.
• Previous models did not fully capture the effects of light-dependent noise and spatial integration limits.

Purpose of the Study:

• To extend the Rovamo, Luntinen & Näsänen (1993b) contrast detection model to low light levels.
• To incorporate the effects of light-dependent quantal noise into the model.
• To validate the extended model by measuring foveal contrast sensitivity across various retinal illuminances, grating areas, and spatial frequencies.

Main Methods:

• Extended a human vision contrast detection model to include light-dependent quantal noise.
• Incorporated optical and neural filtering, internal neural noise, and a local matched filter.
• Measured foveal contrast sensitivity as a function of retinal illuminance and grating area for spatial frequencies from 0.125 to 32 c/deg.

Main Results:

• Contrast sensitivity increased with retinal illuminance (I) and grating area (A) until critical levels were reached, after which it saturated.
• Critical illuminance increased with spatial frequency squared.
• Critical area varied with spatial frequency, decreasing at higher frequencies.
• Maximum contrast sensitivity showed complex dependencies on spatial frequency due to optical and neural factors.

Conclusions:

• The extended model accurately predicts human contrast sensitivity at low light levels, accounting for quantal noise.
• The model successfully explains the observed saturation of contrast sensitivity with increasing retinal illuminance and grating area.
• The findings highlight the interplay between optical limitations, neural processing, and noise in determining visual performance.