As more and more automation is used to manufacture large components, relying on human beings and a set of rudimentary measuring tools to inspect works in progress (WIPs) seems more and more antiquated. More important, manual inspection on automated manufacturing floors now represents the bottleneck of bottlenecks for industries that are otherwise achieving new speed and precision through automation. Laser trackers and coordinate measuring machines (CMMs) are not much faster and don’t measure qualitative features, like the presence of foreign object debris (FOD). Profilometers fulfill only a limited number of inspection applications.
The most promising systems by far for 3D automatic inspection of large WIPs are those that use machine vision. At Aligned Vision, we believe that the next major breakthrough in large component manufacture depends on implementing fast, reliable machine vision-based automatic inspection. And one of the things 3D automatic inspection critically depends on is capturing good data.
Why is good data such a crucial need for 3D automatic inspection of large components? Inspection comprises the recognition and measurement of attributes and the detection of anomalies. In machine vision systems, these functions require analysis algorithms built on statistical processing of captured images. The greater the quantity and level of detail captured, the more refined the statistical results and the more accurate the algorithms and inspection results.
What is good inspection data?
The quality of visual data captured by automatic inspection systems like LASERVISION can be measured in terms of two parameters:
- Pixel density – the number pixels in an image per unit area of the surface being inspected
- Pixel intensity range – the contrast between light and dark pixels, which clarifies light and dark details in the inspected surface
If you look at an eye chart, your ability to distinguish between a 6 and an 8 depends on the level of detail and the amount of light your eye is able to capture. Likewise an automatic inspection system’s ability to distinguish between a black surface and a translucent piece of peel ply depends on the details and light captured by the machine vision camera.
What a 3D automatic inspection system needs to capture good data
What does all this mean for 3D automatic inspection systems applied to large WIPs? To achieve high pixel density and a large pixel intensity range – that is, good data – the machine vision system should offer:
- A large field of view (FOV) so that it can inspect detailed regions of interest anywhere on large, complex surfaces
- A sizeable depth of field to keep features in focus when they are relatively large or involve complex, deep curvatures
- A large standoff distance, the distance from the system’s camera to the WIP surface, needed to achieve a sizeable depth of field
- The ability to capture sufficient contrast in the images under ambient lighting, so that cost-prohibitive special illumination is not required
Finally, the automatic inspection system must offer ultra-high pixel resolution. This is not the same thing as camera resolution, measured in megapixels. Pixel resolution is the number of image pixels within a given region of interest on the WIP surface, i.e. pixel density.
Another way to understand good data in inspection images is to think about the pixelating effect, which obscures details in an image. You would have to magnify an ultra-high resolution image a lot before it would start to pixelate.
Capturing ultra-high resolution images including large depths of field anywhere on large 3D surfaces in ambient lighting – these are the keys to getting good data for automatic inspection. And they are key objectives we have achieved with LASERVISION. To learn more about the LASERVISION system, click here.