Automatic Optical Inspection (AOI) is a non-contact method that uses machine vision technology to analyze captured images to determine whether semi-finished products are defective or not. This inspection method is widely adopted by the industry.
 

       Due to high yield requirements in industries such as the PCB (printed circuit board) industry, the sensitivity of the AOI equipment tends to generate an excessive screening phenomenon. Because of this, a more intelligent inspection system is required. This has led the industry to apply AI technology to assist AOI equipment to optimize the subsequent screening process.
 

       Using the PCB industry as an example, the AOI inspection system primarily employs the optical instrument to scan the PCB to obtain images. It then further compares the qualified parameters in the database and checks for any defects using computer image processing technology. The types of defects that can be detected include short circuits, excessive/insufficient copper, open circuits, notches, burrs, copper slag, missing parts, skewness, etc.

 

       The industry has however pointed out that AOI inspection exhibits an excessive screening phenomenon. The reasons for this can be attributed to the very strict AOI parameter settings in place to cope with the high yield requirements of the PCB industry, and the tendency of light and shadow interferences to cause the AOI to have a sensitive optical principle. Therefore, as long as there is a slight external light and shadow, the device will automatically interpret it as a defective product.
 

       According to industry statistics, when the PCB industry uses AOI equipment for inspection, the probability of finding defects during the screening process is as high as 70%. Because of this, the current industry practice is to perform a second manual screening and once again return the actual qualified PCB boards to the production line. It is estimated that often 4 personnel need to be allocated to an AOI inspection machine for re-screening.
       

       To reduce unnecessary manpower requirements and speed up the inspection on production lines, the PCB industry is currently trying to introduce AI identification modules and using visual identification technology to assist in the subsequent optimization of AOI inspections, to improve the recognition accuracy of the inspection equipment. The PCB industry estimates that the introduction of AI visual identification will effectively reduce the probability of screening to 25%. The difference between AOI inspection and AI identification is whether or not it can identify unknown defects. (News source: DIGITIMES)