Calligraphy and Painting Identification 3D-CNN Model Based on Hyperspectral Image MNF Dimensionality Reduction

Figure 11

Output comparison of the convolution layer of different functional 3D-CNN networks with the same painting and calligraphy data as input. (a) Input layer: first nine channels of MNF images (the previous figure shows the 3D-CNN author classification result/the following figure shows the 3D-CNN authenticity identification result). (b) Output feature images of the fourth convolution layer (the previous figure shows the 3D-CNN author classification result/the following figure shows the 3D-CNN authenticity identification result).

(a)

(b)