Author Affiliations
School of Information and Control Engineering, Xi'an University of Architecture and Technology, Xi'an, Shaanxi 710055, Chinashow less
Fig. 1. Basic structure of CNN
Fig. 2. Convolution process
Fig. 3. Designed CNN model
Fig. 4. Principles of dropout. (a) Network without dropout; (b) network with dropout
Fig. 5. Multispectral images of the pigment true silver
Fig. 6. Flow chart of spectral feature reorganization
Fig. 7. Flow chart of classification experiment for mural pigments
Fig. 8. Multispectral images of standard mural paint board
Fig. 9. Multispectral images of simulated mural
Fig. 10. Sample units after spectral feature recombination of standard pigment
Fig. 11. Classification renderings of different models. (a) Original mural; (b) statistical manifold-SVM model; (c) CNN model (without dropout); (d) CNN model (with dropout)
Pigment type | Mercuric sulfide | Coal black | Tetra green | First green | Lazurite | Minium | Chrome yellow | Gypsum |
---|
Sample No. | 54612 | 64126 | 11705 | 38776 | 70282 | 1155 | 287858 | 882586 |
|
Table 1. Number of samples of each pigment
Category | Mercuric sulfide | Coal black | Tetra green | First green | Lazurite | Minium | Chrome yellow | Gypsum |
---|
Mercuric sulfide | 93.48 | 0 | 0 | 0.75 | 1.72 | 3.65 | 0 | 0 | Coal black | 0 | 81.13 | 0 | 2.95 | 15.52 | 0 | 0 | 0 | Tetra green | 0 | 0 | 90.37 | 9.63 | 0 | 0 | 0 | 0 | First green | 0 | 0 | 29.26 | 70.74 | 0 | 0 | 0 | 0 | Lazurite | 2.16 | 1.38 | 0 | 5.87 | 90.59 | 0 | 0 | 0 | Minium | 13.04 | 0 | 0 | 0 | 0 | 86.96 | 0 | 0 | Chrome yellow | 3.67 | 0 | 0 | 0 | 1.12 | 0 | 95.21 | 0 | Gypsum | 0 | 0 | 0 | 3.02 | 6.24 | 0 | 2.52 | 88.22 |
|
Table 2. Confusion matrix of classification effect of CNN model (with dropout)%
Category | SVM | CNN(no-dropout) | CNN(dropout) |
---|
Mercuric sulfide | 93.37 | 93.49 | 93.48 | Coal black | 80.01 | 80.15 | 81.13 | Tetra green | 90.42 | 90.35 | 90.37 | First green | 67.65 | 70.71 | 70.74 | Lazurite | 86.21 | 90.57 | 90.59 | Minium | 63.78 | 89.92 | 86.96 | Chrome yellow | 88.02 | 95.18 | 95.21 | Gypsum | 84.36 | 88.06 | 88.22 |
|
Table 3. Comparison of classification accuracy of each pigment for three models%