Take your surface inspection to the next level, with automated deep learning artificial intelligence (ai) cosmetic inspection. Checks surfaces, products and components at speed with the knowledge that the more we see, the better we get!
Machine vision surface inspection is used to autonomously confirm the visual and cosmetic quality of products. Machine vision has the advantage of being non-contact, which means it does not contaminate or damage the part being inspected. IVS machine vision systems can inspect for surface deviation, cosmetic changes, visual distortion and surface contamination.
Faster quality checks for cosmetic defects, reducing inspection time, eliminating human-error, without marking your products.
Defects are measured, categorised and data stored. A defect “map” provides real-world data on problem areas on parts and statistical feedback.
Our Deep Learning Artificial Intelligence solutions for surface inspection advance the more they see, improving as it learns.
Every operator and quality audit engineer would like to think they can spot a surface defect when presented with a failed part. But what if the defect were close to the margins of what a human can see or that the product needed to be under certain lighting conditions to view the failure? This is what every manufacturer is up against. And combined with that is the fact that production lines and web produced manufacturing run at speeds which doesn’t allow human inspectors, it’s simply too fast. Operators miss failures either through poor eyesight, repetitive work, tiredness or fatigue – or simply because the defect only shows through certain lighting conditions or angles.
By employing an automated surface inspection system a number of benefits come sharply into focus. Machine vision quality control for surface inspection provides:
As manufacturing processes increase in speed and quality standards become more rigorous, the need for an automated quality control solution is ever more pressing for cosmetic, surface and foreign body detection.
For surface inspection, foreign body detection and cosmetic defect analysis a number of lighting, filter and optic arrangements are normally needed.
Absorbing – Backlighting is used to create contrast of particulates, bubbles and cosmetic defects.
Reflecting – Illumination on-axis to the product creates reflections of fragments, oils and crystallisation.
Scattering – Low angle lighting is used to highlight defects such as cracks, and foreign body fragments.
Polarised – Polarized light is used for highlighting fibers & impurities.
These techniques are normally combined to allow the automated detection of a whole host of surface and cosmetic defect detection, including:
Impurities, particulates, fibers, particles, bubbles, white marks, black inclusions, cracks, scratches, fragments, general marks, pitting, cracking, lumps, dents and material damage.
For automated web inspection a number of linescan cameras are combined capturing the complete width of web to allow finite inspection of the continuous web process at speed. Surface inspection lighting techniques are combined with a single or multi-camera station, coupled with the defect map creation. For medical device inspection multi cameras are combined or the product rotated to allow the completed 360 degree inspection using machine vision for surface anomalies.
Surface inspection for cosmetic, physical and aesthetic faults has tangible benefits across the whole product process.
For example, most web-based production are watched by operators in a patrolling fashion whilst they also attend to set up and maintenance on individual machines. As cost and performance pressures drive up the ratio of machines per operator, less time is available for quality control. This means that defects are going unnoticed, and sometimes a fault is not picked up until it reaches the customer. By employing automated surface inspection vision systems either in-line or at the end of line as part of final inspection, a customer can have the following benefits.
An automated surface inspection solution often replaces an inspector who would have historically been used for surface inspection by eye. But why? Well, there are several limitations to using the old-fashioned way of inspection.
Manual inspection necessitates the presence of a person, an inspector, who assesses the entity in question and renders a decision based on some training or prior knowledge. Except for the trained inspector’s naked eye, no equipment is required.
According to research, visual inspection errors typically range between 20% and 30%. (Drury & Fox, 1975). Some flaws can be attributed to human error, while others are due to space constraints. Certain errors can be reduced but not eliminated through training and practice.
In production processes, visual inspection errors can take one of two forms: missing an existing defect or incorrectly identifying a defect that does not exist (false positive). Misses occur far more frequently than false errors. Misses can result in quality loss, while false positives can result in unnecessary production costs and overall yield reduction.
Human vision assessment is untrustworthy — The human eye’s proclivity to be fooled by optical illusions demonstrates how untrustworthy it can be. This is not to say that manual inspection is completely useless, but it would be unwise to rely solely on it.
Eyesight imperfection — The human eye is incapable of making precise measurements, particularly on a very small scale. Even when comparing two similar objects, the eye may overlook the fact that one is slightly smaller or larger than the other. This concept also applies to surface roughness, size, and any other factor that needs to be measured, especially relating to surface inspection assessments.
Operator costs – Manual inspection is still an expensive due to the appointment of (multiple) trained quality inspection operators.
By making the entire visual inspection procedure independent of any human involvement, automated visual inspection can overcome these issues. Using automated systems typically outperforms manual inspection.
Although machine vision systems can tolerate some variation in the appearance of a part due to scaling, rotation, and pose distortion, complex surface textures and image quality issues pose significant inspection challenges. This is where deep learning artificial intelligence can be applied for surface inspection.
Deep learning-based systems are suitable for more complex surface inspection requirements, such as patterns that vary in subtle but unacceptable ways. Deep learning is effective at learning complex surface and cosmetic defects, such as scratches and dents on turned, brushed, or shiny parts. Deep learning-based image processing, whether used to locate, read, inspect, or classify features of interest, differs from traditional machine vision in its ability to conceptualise and generalise a components overall appearance.
Machine vision performs well at the quantitative measurement of a highly structured scene with a consistent camera resolution, optics and lighting. Deep learning can handle defect variations that require an understanding of the tolerable deviations from the control medium; for example, where there are changes in texture, lighting, shading or distortion in the image. Our deep learning vision systems can be used in surface inspection, object recognition, component detection and part identification. AI deep learning helps in situations where traditional machine vision may struggle, such as parts with varying size, shape, contrast and brightness due to production and process constraints.
Deep Learning in the context of artificial intelligence in machine vision surface inspection is a critical application for the future of manufacturing. IVS® solutions are now developing with this new technology to solve manufacturing inspection tasks which used to be too complicated, time-consuming and costly based on traditional machine vision.