. Author manuscript; available in PMC: 2023 Apr 1.

Published in final edited form as: IEEE Trans Pattern Anal Mach Intell. 2022 Mar 4;44(4):1934–1948. doi: 10.1109/TPAMI.2020.3033882

TABLE 2. Average Metrics on Display Captured PhlatCam measurements.

FlatNet-gen produces higher quality results without compromising on the inference time for both the real PSF case (FlatNet-gen-C) and the simulated PSF case (FlatNet-gen-UC). Le-ADMM shows larger difference in quality between the real and simulated PSF cases owing to its stronger dependence on the PSF.

Method	PSNR (in dB)	SSIM	LPIPS	Inference Time (in sec)
Tikhonov	12.67	0.25	0.758	0.03
TV-ADMM	13.51	0.26	0.755	180
Le-ADMM-UC	18.35	0.49	0.407	0.08
Le-ADMM-C	20.29	0.51	0.333	0.08
FlatNet-gen-UC	20.53	0.54	0.318	0.03
FlatNet-gen-C	20.94	0.55	0.296	0.03