Spatial and spatial-frequency analysis in visual optics

Area of Science:

• Optics
• Image Processing
• Computational Neuroscience

Background:

• Spatial distribution of energy and its Fourier transform (spatial-frequency spectrum) are equivalent for visual targets under linear constraints.
• Power spectrum, lacking phase information, is insufficient for reconstituting the original object.
• Joint representations in space and spatial-frequency dimensions offer a more comprehensive descriptor.

Purpose of the Study:

• To explore the properties of joint representations in space and spatial-frequency dimensions.
• To investigate alternative kernel functions beyond Fourier transforms for visual target analysis.
• To provide a framework for understanding visual information processing.

Main Methods:

• Analysis of joint representations for sample targets using Gabor, Difference-of-Gaussians, and Wigner transforms.
• Substitution of kernel functions (e.g., Gaussian, its derivative) in Fourier transforms.
• Generation of graphs displaying joint properties of targets and their representations.

Main Results:

• Demonstration of the utility of joint representations in characterizing visual targets.
• Exploration of how different transforms and kernels affect the joint representation.
• Development of methods to jointly display target and representation properties.

Conclusions:

• Joint representations are valuable for matching optical device performance to the eye.
• This approach has applications in optical superresolution.
• It aids in analyzing neural processing demands, particularly in visual hyperacuity.