A computer vision system for the inspection of light emitting diodes

This paper outlines the development of a computer vision system to enhance classification criteria for the inspection of infrared light emitting diodes (LEDs). Infrared LEDs used for fibre optics and optical switches are presently classified solely on the basis of power output. This test is performed by focusing the light beam from the LED through a circular aperture onto a solar cell. The output current from the solar cell is translated into a power output reading for the LED. The approach is limited in that it provides no information about other characteristics such as the misalignment of the beam from the mechanical centre or the intensity distribution of the beam. Improved classification criteria and testing methods based on emitted light intensity distribution can be utilised in engineering revision, setting specifications, monitoring manufacturing problems and classification for different applications.