Automating IQI Penetrometer Standard Detection and Measuring the Hole Size

Overview

In the world of Non-Destructive Testing (NDT), accurate interpretation of radiographic images is critical. One of the key elements in evaluating image quality is the Image Quality Indicator (IQI), specifically penetrometers, which help verify the resolution and sensitivity of radiographs.

At SG2 Technologies, we’ve built an intelligent and automated solution for detecting IQI holes using image processing and AI. This solution streamlines inspection workflows, reduces human error, and provides consistent results — all in real time.

Key Features of Our Solution

• Automatic detection of IQI holes using the Hough Circle Transform
• Matching detected holes with known IQI sizes (e.g., 0.241 mm to 8.0 mm)
• Filtering out noise and defects that are not IQI-related
• Identifying the smallest visible IQI hole for sensitivity analysis
• Visualizing and annotating results on the original radiograph

Problem Statement

Manual detection and interpretation of IQI holes on radiographic films can be:

• Subjective: Different inspectors may interpret the visibility of holes differently
• Time-consuming: Especially when reviewing large volumes of radiographs
• Error-prone: Especially in complex or low-contrast images

This led SG2 Technologies to design a fully automated detection pipeline for IQI penetrometers to standardize and accelerate radiographic quality assessment.

How We Did It

1. Preprocessing Radiographic Images

To ensure robustness across varying image qualities and lighting conditions, we applied:

• Contrast Limited Adaptive Histogram Equalization (CLAHE)
• Median filtering to remove salt-and-pepper noise
• Adaptive thresholding to enhance circular features

2. Circle Detection Using Hough Transform

We leveraged OpenCV’s Hough Circle Transform, fine-tuned to detect high-contrast circular features corresponding to IQI holes:

cv2.HoughCircles(image,

                 method=cv2.HOUGH_GRADIENT,

                 dp=1.2,

                 minDist=10,

                 param1=50,

                 param2=30,

                 minRadius=3,

                 maxRadius=20)

3. Matching to Known IQI Hole Sizes

Detected circles were compared to a calibrated reference of IQI hole diameters as per ASTM E1025 and EN 462 standards.

• We applied a tolerance threshold to match detected diameters to standard hole sizes
• Any hole size falling outside expected ranges was discarded

4. Filtering Out False Positives

Using heuristics based on:

• Hole alignment (typically linear in IQIs)
• Relative spacing between holes
• Shape circularity and contrast

…we were able to discard random noise, scratches, or other non-IQI artifacts.

5. Identifying the Smallest Visible Hole

From the valid matches, we programmatically selected the smallest visible IQI hole — a key parameter in sensitivity checks during radiography evaluation.

6. Annotated Visualization

The final output includes:

• Circles drawn over detected holes
• Text annotations with their estimated size in millimetres
• Bounding boxes around the IQI section for context

 Impact

Integration Ready

Our IQI hole detection module is:

• Packaged as a microservice
• Can run on edge devices or cloud
• Provides API endpoints for real-time image upload and analysis
• Compatible with DICONDE, PNG, TIFF formats

Future Enhancements

• Integrate with AI classifiers to identify penetrometer types (wire vs. hole)
• Use deep learning-based segmentation to improve accuracy in complex backgrounds
• Incorporate real-time feedback loop from radiographers to continually improve detection

Conclusion

SG2 Technologies is proud to offer an automated, accurate, and fast solution for IQI penetrometer hole detection. By leveraging advanced computer vision techniques like the Hough Circle Transform and intelligent heuristics, we’ve significantly improved the consistency and speed of radiographic evaluations in NDT workflows.