TY - GEN
T1 - Lightless Fields
T2 - 15th International Symposium on Visual Computing 2020
AU - Vogt, Carson
AU - Lyu, Geng
AU - Subr, Kartic
PY - 2020/12/7
Y1 - 2020/12/7
N2 - Modern focused light field cameras are capable of capturing video at over 160 frames per second, but in so doing sacrifice shutter speed. Outside of laboratory environments, lighting can be problematic resulting in noisy light fields and poor depth reconstruction. To enhance and denoise modern focused light field cameras, we create a unique deep neural network that allows for the full light field to be processed at once, eliminates stitching artifacts, and takes advantage of feature redundancy between neighboring microlenses. We show that our double U-Net network, ENH-W, significantly outperforms several popular architectures and light field denoising methods in both visual and depth metrics.
AB - Modern focused light field cameras are capable of capturing video at over 160 frames per second, but in so doing sacrifice shutter speed. Outside of laboratory environments, lighting can be problematic resulting in noisy light fields and poor depth reconstruction. To enhance and denoise modern focused light field cameras, we create a unique deep neural network that allows for the full light field to be processed at once, eliminates stitching artifacts, and takes advantage of feature redundancy between neighboring microlenses. We show that our double U-Net network, ENH-W, significantly outperforms several popular architectures and light field denoising methods in both visual and depth metrics.
KW - Deep learning
KW - Focused light field camera
KW - Low light
UR - http://www.scopus.com/inward/record.url?scp=85098178777&partnerID=8YFLogxK
U2 - 10.1007/978-3-030-64556-4_30
DO - 10.1007/978-3-030-64556-4_30
M3 - Conference contribution
AN - SCOPUS:85098178777
SN - 9783030645557
T3 - Lecture Notes in Computer Science
SP - 383
EP - 396
BT - Advances in Visual Computing. ISVC 2020
PB - Springer
Y2 - 5 October 2020 through 7 October 2020
ER -