Contrast-to-gradient method for the evaluation of image resolution in scanning electron microscopy

Area of Science:

• Image Processing
• Computational Imaging
• Optics

Background:

• The contrast-to-gradient (CG) method was previously proposed for image resolution evaluation.
• CG resolution is defined as a weighted harmonic mean of local resolution, based on threshold contrast and local gradient.
• Local gradients are derived from quadratic fitting of pixel intensities within a region of interest (ROI).

Purpose of the Study:

• To refine the CG method for more comprehensive image resolution assessment.
• To introduce directional resolutions for evaluating anisotropic image features like astigmatism.
• To ensure CG resolution invariance to contrast reversal and brightness/contrast changes.

Main Methods:

• Incorporation of directional resolutions to analyze images with astigmatism or directional patterns.
• Redefinition of CG resolution for contrast-invariance, maintaining value despite black-and-white reversal.
• Characterization of CG resolution independence from brightness/contrast adjustments within intensity limits.

Main Results:

• The modified CG method provides directional resolution metrics.
• The redefined CG resolution remains consistent for reversed contrast images.
• CG resolution demonstrates robustness against moderate brightness/contrast variations.
• The study analyzes the impact of ROI size on denoising and resolution accuracy relative to image noise.

Conclusions:

• The refined CG method offers a more robust and versatile approach to image resolution evaluation.
• Directional analysis expands the applicability of CG resolution to anisotropic image features.
• Invariance properties enhance the reliability of CG resolution measurements in diverse imaging conditions.