Intelligent evaluation of grotto surface weathering based on spectral chromatic aberration and principal component feature fusion

Chi-peng CAO; Hui-qin WANG; Ke WANG; Zhan WANG; Gang ZHANG; Tao MA

doi:10.37188/OPE.20212910.2444

Journals >Optics and Precision Engineering >Volume 29 >Issue 10 >Page 2444 > Article

Optics and Precision Engineering
Vol. 29, Issue 10, 2444 (2021)

Intelligent evaluation of grotto surface weathering based on spectral chromatic aberration and principal component feature fusion

Chi-peng CAO¹, Hui-qin WANG^1,*, Ke WANG¹, Zhan WANG²..., Gang ZHANG² and Tao MA²|Show fewer author(s)

Author Affiliations

¹School of Information and Control Engineering， Xi'an University of Architecture and Technology， Xi'an70055， China

²Shanxi Provincial Institute of Cultural Relics Protection， Xi’an710075， China

show less

DOI: 10.37188/OPE.20212910.2444 Cite this Article

Chi-peng CAO, Hui-qin WANG, Ke WANG, Zhan WANG, Gang ZHANG, Tao MA. Intelligent evaluation of grotto surface weathering based on spectral chromatic aberration and principal component feature fusion[J]. Optics and Precision Engineering, 2021, 29(10): 2444 Copy Citation Text

show less

Fig. 1. Reflection spectrum characterization of different weathering types and degrees on the surface of grotto

Download full size | View in the Article

Fig. 2. Multi-spectral imaging data of weathering area on the surface of grotto

Download full size | View in the Article

Fig. 3. Technical block diagram of intelligent evaluation method for grotto surface weathering

Download full size | View in the Article

Fig. 4. Reconstructed spectral characteristic curve of grotto surface

Download full size | View in the Article

Fig. 5. Principal component analysis results of multi-spectral image of cave surface weathering

Download full size | View in the Article

Fig. 6. Principal component analysis results image color image image

Download full size | View in the Article

Fig. 7. Overall evaluation results of weathering on the surface of grotto

Download full size | View in the Article

Weathering type	$W_{s t r o n g 1}$	$W_{s t r o n g 2}$	$W_{w e a k 1}$	$W_{w e a k 2}$	$W_{s l i g h t l y 1}$	$W_{s l i g h t l y 2}$	$W_{d u s t 1}$	$W_{d u s t 2}$
L*	46.4	43.7	42.6	37.7	33.0	28.0	8.0	7.9
a*	3.5	4.7	3.5	3.1	3.6	2.5	0.2	0.2
b*	8.9	10.5	8.3	8.7	9.0	7.6	7.9	7.8
X	15.7	13.9	13.0	10.0	7.7	5.5	0.8	0.8
Y	15.6	13.6	12.9	9.9	7.5	5.5	0.9	0.9
Z	9.8	8.1	8.1	6.0	4.4	3.2	0.3	0.3
R	121	117	111	98	88	74	27	27
G	108	100	98	87	75	64	23	23
B	95	85	86	74	63	55	10	10

Table 1. Changes of L*a*b， XYZ and RGB in different weathering types and weathering degree areas

View in the Article

$Δ E$	Chromatic aberration degree identification
0~0.5	Trace
0.5~1.5	Lightweight
1.5~3.0	Can feel
3.0~6.0	obvious
6.0~12.0	tremendous
$\geq$ 12.0	huge

Table 2. Color difference judgment standard

View in the Article

Color difference type	$W_{s t r o n g} - W_{b e n c h m a r k}$	$W_{w e a k} - W_{b e n c h m a r k}$	$W_{s l i g h t l y} - W_{b e n c h m a r k}$	$W_{d u s t} - W_{b e n c h m a r k}$
$Δ E$ 76	45.19	38.81	29.12	11.24
$Δ E$ 94	45.19	38.81	29.12	11.24
$Δ E$ 2 000	32.37	27.01	19.57	8.2

Table 3. Color difference between different weathering types and weathering degrees and reference points

View in the Article

Color difference type	$W_{s t r o n g}$	$W_{w e a k}$	$W_{s l i g h t l y}$	$W_{d u s t}$	$W_{b e n c h m a r k}$
$W_{s t r o n g}$	0	5.83	12.68	21.01	74.05
$W_{w e a k}$	5.83	0	6.85	5.83	68.22
$W_{s l i g h t l y}$	12.68	6.85	0	8.33	61.37
$W_{d u s t}$	21.01	15.18	8.33	0	53.04
$W_{b e n c h m a r k}$	74.05	68.22	61.37	53.04	0

Table 4. Color difference between different weathering types and degrees

View in the Article

Principal component	Eigenvalues	Contribution rate	Cumulative contribution rate
PC1	16757.90	95.41%	95.41%
PC2	579.80	3.30%	98.71%
PC3	92.21	0.53%	99.24%

Table 5. Contribution rate and cumulative contribution rate of the first three principal components

View in the Article

Method	Accuracy evaluation	Color difference-principal component	Reflectance spectrum	principal component	Color difference
RF	Training accuracy	100.00%	99.91%	98.52%	93.03%
	Prediction accuracy	99.86%	98.49%	91.34%	70.26%
	Kappa coefficient	0.99	0.98	0.89	0.53
KNN	Training accuracy	99.79%	94.99%	88.60%	75.43%
	Prediction accuracy	97.50%	91.69%	82.68%	66.84%
	Kappa coefficient	0.96	0.90	0.69	0.42
BP	Training accuracy	63.93%	63.19%	46.13%	36.84%
	Prediction accuracy	50.81%	48.00%	42.19%	32.51%
	Kappa coefficient	0.28	0.25	0.21	0.18
RBF	Training accuracy	96.12%	95.95%	87.52%	70.08%
	Prediction accuracy	94.73%	92.88%	85.96%	65.84%
	Kappa coefficient	0.92	0.91	0.75	0.39

Table 6. Comparison of accuracy and kappa coefficient of four evaluation methods

Download Citation

Save the article for my favorites

Paper Information

微信扫一扫：分享

微信扫一扫：分享