Machine learning based luminance analysis of a $μ$LED array

Abstract: In the past years, the development of $\mu$LED arrays gained momentum since they combine the advantages of $\mu$LEDs, such as high brightness and longevity, with a high resolution of a micro-scaled structure. For the development, spatially resolved measurements of luminance and color of single $\mu$LEDs and the entire light-emitting surface are analyzed as they are crucial for the visual perception. However, the former is time intense in measurement and evaluation, and the latter suffers from interference caused by nonfunctional $\mu$LEDs. This paper presents a method to perform both analyzes with a single measurement using unsupervised machine learning. The results suggest that a precious reconstruction of the $\mu$LEDs and a more accurate characterization $\mu$LED arrays can be achieved.