Human sensitivity to phase perturbations in natural images: a statistical framework.

Area of Science:

• Visual perception
• Computational neuroscience
• Image processing

Background:

• Fourier-phase information is crucial for natural scene appearance but complex to link to human perception.
• Previous studies explored human visual sensitivity to phase randomization and quantization in natural images.

Purpose of the Study:

• To investigate how human visual sensitivity relates to Fourier phase manipulation in natural images.
• To determine the perceptual equivalence of phase randomization and quantization processes.
• To constrain models of cortical processing based on visual sensitivity to phase changes.

Main Methods:

• Analyzing human visual sensitivity to randomized and quantized Fourier phase spectra of natural images.
• Quantifying the salience of image changes induced by phase manipulation.
• Comparing perceptual effects of randomization and quantization based on average phase shifts.

Main Results:

• The impact of phase randomization and quantization on image salience critically depends on the original phase spectrum.
• Phase randomization and quantization are perceptually equivalent if they induce the same average phase shift.
• Visual sensitivity is linked to phase-domain changes that alter global higher-order image statistics.

Conclusions:

• The visual system detects phase-domain changes that affect higher-order image statistics.
• Cortical processing likely involves mechanisms correlating outputs of relative-phase-sensitive units.
• This provides insights into the neural basis of visual processing for natural scenes.