Industrial inspection is critical for 3D semiconductor manufacturing process and automated optical inspection for 3D ICs has attracted a lot of attentions in these years. Plenoptic imaging systems, based on a micro-lens array, acquire light field from parallax and computes 3D information with lower costs. To reduce aberrations from the optical design with microlens array, especially for off-axial micro lenses, the design flow for plenoptic imaging systems is proposed. Based on the parameters designed from paraxial approximation, lateral image quality is optimized by a commercial optical design software, and then depth-related performances are estimated from the simulated images of the optical system. The experimental system for validation is tested quantitatively with modulation transfer function (MTF), by the slanted-edge method of ISO 12233. The difference of MTF between the paraxial and off-axial regions is approximately 0.02, which is within the repeatability error 0.03. Moreover, the synthesized images of a PCIe card refocused on the chip and the board clearly show the elements at the refocusing depth only. The depth map and the all-in-focus image are estimated to build a 3D model. However, significant artifacts appear on depth maps when lighting is not uniform. With combination of the ring light and coaxial light, the depth maps of objects with different surface properties can be estimated with less artifacts. Furthermore, accuracy and resolution can be enhanced by deep-learning technologies, which will be implemented in the future.
Optical design for reducing aberrations of the micro-lens-array-based
integral imaging system is challenging. A design process combining the
sequential and non-sequential modes of optical design software is
proposed. The process is verified by a system assembled on a
coordinate measuring machine with errors of several micrometers.
Differences in the modulation transfer function, measured by the
slanted-edge method, are less than 0.02 between the paraxial and
off-axial regions. Reconstructed images of a U.S. quarter-dollar coin
with different refocusing depths show the synthesized defocusing. The
estimated depth map and depth-based reconstructed image show the rough
shape of the coin.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.