Fig. 6: Frequency up-conversion parity classifier of handwritten numbers. | Nature Communications

Fig. 6: Frequency up-conversion parity classifier of handwritten numbers.

From: Laser nanoprinting of 3D nonlinear holograms beyond 25000 pixels-per-inch for inter-wavelength-band information processing

Fig. 6

a The experimental diagram. We use a 1342 nm light to illuminate the number and combine it with a 1064 nm light. Then, we input these NIR lights on the designed nonlinear hologram. The output SF light at 593 nm is focused onto different positions according to the number parity. b The phase distribution of nonlinear hologram. c and d shows the confusion matrices after testing 1000 and 100 handwritten numbers in simulation and experiment, respectively. The diagonal and off-diagonal elements in (c) and (d) represent the correct and incorrect predictions, respectively.

Back to article page