In general terms, optical imaging directly measures the absorption and scattering properties of a component under investigation. High-resolution digital cameras are employed to capture the live images, which are instantly transferred into a computer to conduct digital image processing. By applying a wide range of algorithms, together with a sophisticated computer software system, image segmentation, feature classification, image interpretation, and pattern recognition can be implemented efficiently, automatically or semi-automatically with certain interactive manipulations by the operator.

Optical imaging and digital image processing play an important role in Non-Destructive Testing. There are a number of advantages over other commonly used techniques such as ultrasound and electromagnetic methods. For example, optical imaging can acquire information (i.e. images) of a component without direct contact. It is also an area scanning method, capable of instantaneous recording a two-dimensional image of an object. This is especially important for investigating dynamic behaviours of a component.

Once the live images are taken, appropriate image processing reveals potential defects within the component under investigation. The most commonly used image-processing algorithm is digital image subtraction. Firstly, a Golden Image of a standard, defect-free sample is captured and stored in a computer. Subsequently, live images of components to be inspected are taken.

In order to carry out successive NDT inspections for a stream of components in a production line, a conveyor system may be utilised. This transports components to the right position so that proper optical images can be taken.

For large components, several local images will be taken; each corresponding to part of the component. This is done either by moving the camera or moving the component for local focusing. All local images are then merged to form a complete bigger image. The next step is to compare the live image with a golden reference image to work out differences between them. The residual image left behind will be black except for areas of difference. These are therefore potentially defective. Further image processing involving image segmentation, classification and recognition will finally lead to the detection of potential defects within the components.

Another important image processing method essential for optical NDT technique is image registration. The purpose of image registration is to transform different sets of images into one coordinate system to compensate for misalignement during image capturing under different conditions. Only after image registration can the images obtained from different measurements be compared or integrated directly.

In the automated optical NDT system described above, the component under inspection is expected to be positioned to the right place as precisely as possible by a mechanical positioning system. However, due to the system's precision or other factors, there is often some error in positioning. This makes direct image subtraction impossible. By applying proper image registration, those image misalignments can be compensated for.

The final step for an Automated Optical Imaging (AOI) system is to send a screening signal for each component so that defective components can be removed from the production line. This can be done by setting up an appropriate threshold for each type of defect. When the difference is greater than a predefined threshold, an alarm is signalled and the component is categorised as failed.

The AOI system at The NDT Validation Centre is used for the inspection of PCBs, and is capable of detecting defects such as missing components, dry joints, solder shorts, poor wetting and misaligned components, amongst others. It is also being developed for surface inspection all materials, and is a viable alternative to conventional visual inspection.

By combining AOI with other inspection techniques, such as micro-focus X-ray, acoustic methods, and thermal/infrared imaging, a complete optical inspection system could be established to perform all high technology optical inspections conceivable, with more sophisticated optical inspection procedures being developed.