International Display Workshops General Incorporated Association

15:10 〜 15:30

[DXR2-2] End-to-End Computational Lensless Imaging with Perceptual Loss

*Ya-Ti Chang Lee1、Chung-Hao Tien1 (1.National Yang Ming Chiao Tung University (Taiwan))

Lensless imaging, Perceptual loss, Artificial neural network, Coded aperture

https://doi.org/10.36463/idw.2022.1153

Recently, computational lensless imaging had been making progress with the evolution of artificial neural networks. Nonetheless, generative models for image reconstruction inherit challenge due to its ill-posed nature. We proposed a deep neural network based lensless imaging system by optimizing perceptual loss exclusively to end-to-end reconstruct images conforming human preference.