Automatic Optical Inspection (AOI) or automated optical inspection, is a key equipment used in the quality control of electronics printed circuit boards (PCB) and PCB Assembly (PCBA). Automatic optical inspection check the position of electronic component on PCBs, and the connections between them, to ensure PCBA can work well.

But, only limited people know how Automatic Optical Inspection works, and what is the workflow of AOI. Here, the principle and workflow of AOI is introduced  in details.

The Workflow Chart of Automatic Optical Inspection (AOI)

The Workflow Chart of Automatic Optical Inspection (AOI)

Step1: Image comparison principle 

At first,  CCD camera capture the images of PCB. Then, a specialized intelligent application software processe the CCD singnals.  After that, the image transform into the information we need, such pixel distribution, brightness and color, etc., into digital signals.

Above all, the comparison between the image of tested PCBs and standard image decide if tested PCBS can PASS the AOI check. Normally, the comparision include the size, angle, offset, brightness, color, and position of the component between the tested and standard image.

Step 2: Statistical modeling principle

Automatic Optical Inspection (AOI) statistical modeling enhances system identification OK and NG by learning a series of standard templates.

Through observing image changes and combining the visual deviations in all images of standard templates, AOI will find out the shape changes of components and possible future changes.

In this process, AOI solve three key problems:

A. What should the component look like?
That is, for instance, the size, shape, color, and surface pattern of the component.

B. What changes will happen to the components?
That is, for instance, the natural size, shape, color, and surface pattern of the component,

C. How much will the shape of the component change?
That is, the size, shape, color, surface pattern, and the like of the component are somewhat reasonable.

In a word, the final result is a standard model for testing between OK and NG that combines the above elements.