Texture analysis microscopy: quantifying structure in low-fidelity images of dense fluids

Area of Science:

• Microscopy and Image Analysis
• Materials Science
• Fluid Dynamics

Background:

• Optical images frequently suffer from noise, low contrast, and artifacts, hindering accurate image analysis, especially for dense fluids.
• Conventional methods like global arithmetic operations and kernel-based convolution for noise removal and contrast enhancement have limitations and can degrade image details.

Purpose of the Study:

• To introduce a new technique, texture analysis microscopy (TAM), designed to overcome the limitations of traditional methods in optical image analysis.
• To demonstrate the effectiveness of TAM in enhancing the analysis of low-fidelity images, particularly in challenging conditions.

Main Methods:

• Developed texture analysis microscopy (TAM), a novel technique based on image correlation.
• TAM analyzes images by measuring statistical similarities between the raw image and a template feature (e.g., a Gaussian).
• Applied TAM to low-fidelity optical images where conventional techniques perform poorly.

Main Results:

• TAM successfully improved the quality of noisy and low-contrast optical images.
• Demonstrated superior performance of TAM compared to traditional methods under challenging imaging conditions.
• Enabled accurate structural correlations, particle identification, and sizing in degraded images.

Conclusions:

• Texture analysis microscopy (TAM) offers a robust solution for analyzing challenging optical images, outperforming conventional approaches.
• TAM's image correlation-based approach enhances image fidelity, facilitating detailed analysis of dense fluids and particle characteristics.