Fig. 4: Image feature extraction and recognition using polarization-state-dependent convolutional kernels. | Nature Communications

Fig. 4: Image feature extraction and recognition using polarization-state-dependent convolutional kernels.

From: Multi-dimensional visual information processing under complex light environments using time-evolved polarization-sensitive synaptic electronics

Fig. 4

a Schematic diagram of artificial visual preprocessing using polarization-state-dependent convolutional kernels and a three-layer ANN based on the synapse. b Feature extraction of the input original building image in vertical, horizontal, and edge modes according to the STP-defined edge detection (scale bar: 10 m). c Recognition images of handwritten digit “6” in the MNIST dataset using the STP-defined edge detection (vertical feature extraction) before applying noise and (d) after applying 50% salt pepper noise, respectively. e Equivalent circuit diagram of a 3 × 3 convolution kernel based on a 3 × 4 microwire array and (f) the corresponding outputs after hardware convolution operations. g The recognition accuracy under different information processing conditions. h The confusion matrixes by ANN with convolutional kernels and added noise.

Back to article page