Grain Size Analysis Tutorial

Introduction

• Presenter: Srini
• Channel: Python for Microscopists
• Topic: Real-life application of grain size analysis using Python.

Overview

• Grain Size Analysis: Loading images of grains (or cells for life scientists) to understand size distribution.
• Importance of defining scale in images: Average grain size in pixels vs. microns.

Steps in the Analysis

1. Read Image and Define Pixel Size

   • Import necessary libraries:
     • cv2, numpy, scipy, skimage, matplotlib
   • Read the image using cv2.imread in grayscale.
   • Define conversion factor from pixels to microns (e.g., 1 pixel = 0.5 microns).

2. Clean Up the Image (if needed)

   • Use denoising techniques if the image is noisy (e.g., median filter, non-local means filter).
   • Threshold the image to separate grains from boundaries:
     • Examine histogram to determine threshold value.
     • Apply OTSU's method using cv2.threshold.

3. Erosion and Dilation

   • Erode the thresholded image to reduce noise, followed by dilation to restore size.
   • Define a kernel for operations and perform:
     • cv2.erode
     • cv2.dilate

4. Label the Grains

   • Use ndimage.label to assign labels to identified grains, based on connectivity.
   • Define a structure factor for connectivity (8-connectivity used).

5. Measure the Properties of Each Grain

   • Use measure.regionprops to extract properties:
     • Properties include area, equivalent diameter, orientation, etc.
   • Loop through the clusters to print relevant properties.

6. Output Results to CSV File

   • Write data into a CSV file for further analysis.
   • Structure includes headers for properties and corresponding values for each grain.

Conclusion

• Data can be exported to Excel for analysis, visualization, or further statistical evaluation.
• Future tutorials may focus on different types of images or more advanced analysis techniques.

Call to Action

• Encourage viewers to subscribe and interact with the content.