Defect Detection in Transparent Film Material

Application and Challenge

Traditional image processing often falls short for surface inspection, especially with transparent materials like uncoiled plastic film. Defects such as bubbles, holes, and inclusions can be difficult to distinguish from the background, leading to high rates of pseudo-rejects or, worse, undetected flaws. These issues arise because classical image processing tools, despite extensive options for pre-processing and filtering, struggle with the complex visual characteristics of transparent film.

Solution

To address these challenges, senswork's Neuralyze® software has been employed to implement an AI-driven inspection system specifically designed for transparent film material. This solution integrates advanced techniques, including transmitted light deflectometry, which is essential for making defects visible in the first place.

Transmitted light deflectometry works by directing a strip of light through the film, where diffraction effects caused by material damage create a signal that is easily visible to the human eye. However, standard algorithms typically struggle to detect these signals because the background image often features a sinusoidal pattern, complicating the process.

ROI Calculator

Discover the potential cost savings of robotic automation over a 20-year system life

This calculator compares your current manual labor costs against the total cost of owning and operating a robotic system over its 20-year lifespan.

EXPLORE TODAY

Neuralyze® overcomes this obstacle by leveraging AI to analyze the complex image data. It accurately identifies bubbles, holes, and other defects with a high degree of reliability, even when these defects are subtly embedded within the film's background pattern.

Results

The implemented solution enables effective inspection of transparent surfaces, even in cases where traditional methods struggle to differentiate defects from the background. It achieves high-resolution detection down to 40 µm, ensuring that even very small imperfections are reliably identified. Additionally, the system delivers consistent and dependable results, significantly reducing pseudo-rejects while minimizing the risk of undetected flaws.

To learn more about Vision AI for quality automation, click here.