Meaning
Optical manipulation techniques control the Fourier components of an image to isolate specific surface frequencies during automated quality control. By placing a physical aperture or a digital mask at the Fourier plane of an imaging system, spatial filtering removes high-frequency noise or low-frequency illumination gradients to enhance the contrast of micro-defects. This process is standard in high-speed optical inspection scanners.
Frequency Selection
Low-pass filters remove fine surface roughness and electrical noise from the image to expose larger structural anomalies. High-pass filters do the opposite by isolating sharp edges and scratches. This selective modification simplifies automated image processing.
Contrast Optimization
In semiconductor packaging inspection, the repeated patterns of lead frames and solder bumps generate strong diffraction signals that can obscure micro-cracks in the silicon substrate. By utilizing a custom spatial filter that blocks these periodic diffraction spots in the Fourier plane, the imaging system suppresses the pattern of the circuit and exposes the random defects. This optical pre-processing allows the inspection tool to run at much higher speeds than would be possible if the pattern suppression were executed entirely by software.
The resulting high-contrast images are processed by neural networks to identify defects in real time.
Defect Discernment
The choice between physical masks and digital filtering algorithms depends on the speed requirements of the factory line. Physical apertures placed in the lens system run at the speed of light, making them suitable for ultra-fast web scanners. Digital filters offer greater flexibility when processing multi-layered circuit boards with varying grid designs.