New computational solution to quantify synthetic material porosity from optical microscopic images.

Area of Science:

• Materials Science
• Computer Science
• Image Analysis

Background:

• Accurate quantification of synthetic material porosity is crucial for material characterization and performance prediction.
• Existing methods for porosity analysis from microscopic images can be time-consuming and may lack precision.

Purpose of the Study:

• To develop and evaluate a novel computational solution for automatic porosity quantification in synthetic materials using optical microscopic images.
• To compare the performance of the new solution against human analysis and a commercial system.

Main Methods:

• Development of a computational solution utilizing a multilayer perceptron artificial neural network.
• Training the network using a backpropagation algorithm.
• Validation through analysis of 40 synthetic material images, including noisy samples, and comparison with human assessment and a commercial system.

Main Results:

• The new solution demonstrated superior accuracy and efficiency in quantifying porosity compared to a commercial system.
• Human visual analysis confirmed the high quality of the results obtained by the automated solution.
• The system exhibited robustness and reliability when analyzing images containing noise.
• The training phase was found to be simple and straightforward.

Conclusions:

• The developed artificial neural network-based solution provides an automatic, fast, efficient, and reliable method for quantifying synthetic material porosity from microscopic images.
• This approach is a valuable tool for researchers, engineers, and professionals involved in material science and characterization.