Figure S4. Computational light-field renderings and 3D depth
reconstructions of numeric cube targets by using the fully integrated
NIR-ULFC without NIR bandpass filter. Each of numeric targets are
located at 40 mm, 90 mm, and 180 mm from the NIR-LFC. a) Vertically
shifted sub-aperture images with different viewing angles. The white
dotted lines indicate the original position of numeric cube ‘3’ (center
sub-aperture image), and the images on the left and right are obtained
by changing the viewing angles up and down. b) Digital refocusing
results of the targets, which focus on ‘1’ (left), ‘2’ (center), and ‘3’
(right), respectively. Asterisk symbols at each image indicate the
focused point. c) Disparity map calculated by pixel-by-pixel cost volume
filtering from the whole 27 x 27 sub-aperture images. d) 3D
reconstruction of the targets after image warping by using a disparity
map and a center sub-aperture image. The 2D image is warped on the
disparity map. e) 3D depth map containing discrete depth information is
calculated by using the disparity map obtained from c) and camera
matrices, i.e., intrinsic parameter ([K ]) and extrinsic
parameter ([R |t ]) from geometric calibration
steps. The calculated z-axis depth for each numeric target is 44 mm, 97
mm, and 191 mm, respectively, similar to those of actual positions.