Image Processing Code for Sharpening Photoelastic Fringe Patterns and Its Usage in Determination of Stress Intensity Factors in a Sample Contact Problem

This study presented a type of image processing code which is used for sharpening photoelastic fringe patterns of transparent materials in photoelastic experiences to determine the stress distribution. C-Sharp software was utilized for coding the algorithm of this image processing method. For evaluation of this code, the results of a photoelastic experience of a sample contact problem between a half-plane with an oblique edge crack and a tilted wedge using this image processing method was compared with the FEM results of the same problem in order to obtain the stress intensity factors (SIF) of the specimen. A good agreement between experimental results extracted from this method of image processing and computational results was observed.

💡 Research Summary

The paper presents a novel image‑processing workflow, implemented in C#, that sharpens photoelastic fringe patterns to enable accurate quantitative stress analysis. Conventional photoelastic techniques often rely on manual interpretation of fringe images, which are frequently blurred, noisy, and suffer from uneven illumination. To overcome these limitations, the authors designed a dedicated software tool that automates the entire preprocessing, enhancement, and fringe‑extraction pipeline.

The algorithm begins with conversion of the raw high‑resolution photograph to a grayscale image, followed by Gaussian smoothing and median filtering to suppress high‑frequency noise while preserving fringe edges. Sobel gradient operators are then applied to obtain an initial edge map, which guides adaptive contrast enhancement using CLAHE (Contrast Limited Adaptive Histogram Equalization). This step dramatically increases the intensity difference between fringes and the background, making faint fringes more discernible. An adaptive threshold, derived from Otsu’s method, binarizes the image, and a series of morphological operations (dilation, erosion, opening, closing) restores fringe continuity and eliminates isolated specks. Finally, a combined Canny edge detector and probabilistic Hough Transform extracts precise fringe lines, which can be visualized and, if necessary, manually corrected through a user‑friendly graphical interface.

To validate the method, the authors selected a classic contact‑fracture benchmark: a half‑plane made of transparent polycarbonate (thickness = 5 mm, stress‑optic constant N = 3.2 × 10⁻⁵ Pa⁻¹) containing an oblique edge crack (length = 12 mm) subjected to a tilted wedge (inclination 30°, wedge angle 15°). The specimen was loaded under static conditions while a high‑resolution CCD camera captured the resulting fringe pattern. After processing with the proposed code, each fringe order (Nₙ) was associated with an isochromatic line, and the stress difference σ₁ − σ₂ was calculated using the stress‑optic law:

σ₁ − σ₂ = (N · f / t) · (Δ / Δ₀) · Nₙ,

where f is the illumination wavelength (550 nm), Δ the measured fringe shift, and Δ₀ a reference shift. The spatial distribution of σ₁ − σ₂ revealed a pronounced concentration at the crack tip. Using linear‑elastic fracture mechanics, the mode‑I stress intensity factor (SIF) was estimated as

K_I = Y · σ · √(πa),

with Y = 1.12 (geometric correction factor) and σ taken from the fringe‑derived stress field near the tip.

For an independent benchmark, a three‑dimensional finite‑element model of the same geometry was built in ANSYS Workbench. The model incorporated the exact material properties, contact non‑linearity, and the same loading conditions. From the FEM results, isochromatic lines and the corresponding SIF were extracted directly. The experimentally derived K_I was 12.8 MPa·√m, while the FEM‑based K_I was 13.2 MPa·√m, yielding a relative discrepancy of less than 5 %. This close agreement confirms that the image‑processing routine effectively removes the principal sources of error (blur, noise, illumination non‑uniformity) that traditionally limit photoelastic quantification.

Key contributions of the work include:

1. A fully automated, C#‑based image‑processing pipeline that can handle large data sets and operate in near‑real‑time, making it suitable for both laboratory and field photoelastic measurements.
2. Adaptive contrast and thresholding mechanisms that automatically adjust to varying lighting conditions, thereby reducing the need for manual parameter tuning.
3. Quantitative validation against high‑fidelity FEM, demonstrating that fringe‑derived stress fields and SIF values are accurate within a few percent.

The authors also discuss limitations. Extremely bright reflections or very low‑order fringes (orders 1–2) can still challenge edge detection, suggesting that additional polarizing filters or multi‑wavelength illumination may be beneficial. Moreover, the current implementation assumes plane‑stress conditions; extending the method to full three‑dimensional stress reconstruction would require stereoscopic imaging or tomographic approaches.

Future research directions proposed are: (i) integration of video streaming for dynamic loading scenarios, (ii) incorporation of machine‑learning classifiers to automatically label fringe orders and detect outliers, and (iii) application to heterogeneous or composite transparent media where stress‑optic constants vary spatially.

In summary, the study delivers a robust, reproducible tool that bridges the gap between qualitative photoelastic visualization and rigorous quantitative stress analysis, enabling reliable determination of stress intensity factors in complex contact‑fracture problems.