Information Enhancement Method for Surface Disease Images of Ancient City Walls Based on Adaptive Correction of Illumination Component

Jin Wang; Huiqin Wang; Ke Wang; Zhan Wang

doi:10.3788/LOP202259.1610012

Journals >Laser & Optoelectronics Progress >Volume 59 >Issue 16 >Page 1610012 > Article

Laser & Optoelectronics Progress
Vol. 59, Issue 16, 1610012 (2022)

Information Enhancement Method for Surface Disease Images of Ancient City Walls Based on Adaptive Correction of Illumination Component

Jin Wang¹, Huiqin Wang^1、*, Ke Wang¹, and Zhan Wang²

Author Affiliations

¹College of Information and Control Engineering, Xi’an University of Architecture and Technology, Xi’an 710055, Shaanxi , China

²Shaanxi Provincial Institute of Cultural Relics Protection, Xi’an 710075, Shaanxi , China

show less

DOI: 10.3788/LOP202259.1610012 Cite this Article Set citation alerts

Jin Wang, Huiqin Wang, Ke Wang, Zhan Wang. Information Enhancement Method for Surface Disease Images of Ancient City Walls Based on Adaptive Correction of Illumination Component[J]. Laser & Optoelectronics Progress, 2022, 59(16): 1610012 Copy Citation Text

show less

Changes of image brightness before and after correction under different lighting conditions. (a) 2D-Gamma function; (b) enhanced 2D-Gamma function

Fig. 1. Changes of image brightness before and after correction under different lighting conditions. (a) 2D-Gamma function; (b) enhanced 2D-Gamma function

Download full size

Image comparison before and after the enhanced 2D-Gamma function correction. (a) Before the enhanced 2D-Gamma function correction; (b) after the enhanced 2D-Gamma function correction

Fig. 2. Image comparison before and after the enhanced 2D-Gamma function correction. (a) Before the enhanced 2D-Gamma function correction; (b) after the enhanced 2D-Gamma function correction

Download full size

Fig. 3. Block diagram of homomorphic filtering algorithm

Download full size

Fig. 4. Homomorphic filter transfer function

Download full size

Fig. 5. Comparison of brightness histograms of images before and after correction. (a) Brightness histogram before correction; (b) brightness histogram after correction

Download full size

Fig. 6. Influence of the weight coefficient on the uniformity of illumination

Download full size

Fig. 7. Image comparison before and after linear weighted fusion. (a) Homomorphic filtered image; (b) linearly weighted fusion image

Download full size

Fig. 8. Accuracy of edge extraction operator for disease recognition

Download full size

Fig. 9. Influence of different weighting factors ε on the accuracy of city wall disease recognition

Download full size

Fig. 10. Flow chart of the proposed method

Download full size

Fig. 11. Four algorithm processing results. (a) Original images; (b) Gamma function;（c）homomorphic filtering；（d）Enlighten GAN；（e）proposed method

Download full size

Image brightness v（x，y）	i（x，y）=0	i（x，y）=64	i（x，y）=128	i（x，y）=192	i（x，y）=255
0	0	0	0	0	0
20	134	73	20	1.6	0.004
40	160	102	40	5.7	0.16
60	177	125	60	14	0.83
80	190	144	80	25	2.59
100	201	161	100	39	9.8
120	211	176	120	56	18
140	219	190	140	77	32
160	227	203	160	101	51
180	233	214	180	125	73
200	239	225	200	155	107
220	245	236	220	189	150
240	249	247	240	225	205
255	255	255	255	255	255

Table 1. Image brightness values corrected by enhanced 2D-Gamma function under different lighting conditions

Image	Evaluation index	Original image	Gamma correction	Homomorphic filtering	Enlighten GAN	Proposed method
Image 1	Average illumination	5.2006	5.2126	9.6416	6.2620	6.2450
	Illumination uniformity	0.5988	0.6401	0.3428	0.7903	0.6346
	Image details	32.1773	32.8856	16.7305	29.1454	37.6107
	Average gradient	10.29	10.43	5.31	9.1360	11.93
	Weighted evaluation	12.2843	12.4890	10.1998	11.9734	14.4327
Image 2	Average illumination	5.0787	5.0896	8.4257	5.5420	6.0403
	Illumination uniformity	0.5321	0.5319	0.3719	0.7449	0.5759
	Image details	32.2327	32.8913	17.3015	31.5139	38.8597
	Average gradient	10.19	10.33	5.45	9.7957	12.21
	Weighted evaluation	12.2134	12.4022	9.6940	12.3071	14.6647
Image 3	Average illumination	5.1728	5.1746	8.8502	6.6844	6.3356
	Illumination uniformity	0.6121	0.6119	0.3478	0.8112	0.6475
	Image details	32.7037	33.4407	19.9558	29.3418	38.0983
	Average gradient	10.49	10.65	6.36	9.2131	12.12
	Weighted evaluation	12.4275	12.6335	10.7077	12.2664	14.6300

Table 2. Comparison of performance indicators of four methods

Method	Total number of test images	Correct recognition number	Accuracy /%
Original image	128	96	74.80
Gamma correction	128	105	77.95
Homomorphic filtering	128	108	84.25
Enlighten GAN	128	71	55.47
Proposed method	128	117	91.41

Table 3. Recognition accuracy of four methods

Noise density	Method	Average illumination	Illumination uniformity	Image details	Average gradient	Weighted evaluation
	Gamma correction	5.5107	0.6060	47.8939	15.0239	16.9986
0.02	Homomorphic filtering	9.0385	0.4238	34.2431	10.5913	14.9384
	Enlighten GAN	5.5075	0.6061	47.8612	15.0062	16.9876
	Proposed method	6.7123	0.6085	49.3029	15.3466	18.0551
	Gamma correction	5.4313	0.6078	56.6901	17.6842	19.5028
0.04	Homomorphic filtering	8.8502	0.4220	41.9136	12.8809	17.0500
	Enlighten GAN	5.4270	0.6084	56.6316	17.6889	19.4849
	Proposed method	6.5405	0.6102	57.7359	17.9046	20.4016
	Gamma correction	8.6394	0.4190	47.4667	14.5372	18.5375
0.06	Homomorphic filtering	5.3598	0.6103	63.8462	19.8968	21.5387
	Enlighten GAN	5.3569	0.6121	63.7545	19.9345	21.5193
	Proposed method	6.3608	0.6124	64.7129	20.0575	22.3270
	Gamma correction	5.2989	0.6130	60.8463	21.7687	20.9988
0.08	Homomorphic filtering	8.5363	0.4075	51.2301	15.6547	20.3643
	Enlighten GAN	5.2994	0.6163	69.7201	20.7456	23.2147
	Proposed method	6.2507	0.6085	70.3112	21.7950	23.8916
	Gamma correction	5.2437	0.6156	70.2117	23.4434	23.6277
0.10	Homomorphic filtering	8.4295	0.4049	54.4469	16.6153	20.4363
	Enlighten GAN	5.2486	0.6201	75.0513	23.5530	24.7548
	Proposed method	6.1166	0.6088	75.4713	23.3984	25.3772

Table 4. Results of the evaluation of image quality with noise

Download Citation

Set citation alerts for the article

Tools

Set citation alerts for the article

Save the article for my favorites

Paper Information