Multi-scale Convolution Combined with Adaptive Bi-interval Equalization for Image Enhancement

Hao-xiang LU; Zhen-bing LIU; Peng-yue GUO; Xi-peng PAN

doi:10.3788/gzxb20204910.1010002

Journals >Acta Photonica Sinica >Volume 49 >Issue 10 >Page 1010002 > Article

Acta Photonica Sinica
Vol. 49, Issue 10, 1010002 (2020)

Multi-scale Convolution Combined with Adaptive Bi-interval Equalization for Image Enhancement

Hao-xiang LU¹, Zhen-bing LIU¹, Peng-yue GUO², and Xi-peng PAN¹

Author Affiliations

¹School of Computer and Information Security，Guilin University of Electronic Technology，Guilin，Guangxi 541004，China

²School of Electronic Engineering and Automation，Guilin University of Electronic Technology，Guilin，Guangxi 541004，China

show less

DOI: 10.3788/gzxb20204910.1010002 Cite this Article

Hao-xiang LU, Zhen-bing LIU, Peng-yue GUO, Xi-peng PAN. Multi-scale Convolution Combined with Adaptive Bi-interval Equalization for Image Enhancement[J]. Acta Photonica Sinica, 2020, 49(10): 1010002 Copy Citation Text

show less

Fig. 1. Algorithm flow

Download full size | View in the Article

Fig. 2. Multi-scale convolution extraction process of image details

Download full size | View in the Article

Fig. 3. The process of solving threshold values by genetic algorithm

Download full size | View in the Article

Fig. 4. Image fusion result with different β

Download full size | View in the Article

Fig. 5. Original infrared images of building, enhancement results by different methods and their histograms

Download full size | View in the Article

Fig. 6. The enhancement effect and local detail image of the infrared images of building

Download full size | View in the Article

Fig. 7. Image fusion result with different λ

Download full size | View in the Article

Fig. 8. The enhancement effect of different algorithms

Download full size | View in the Article

Fig. 9. The enhancement effect of different algorithms

Download full size | View in the Article

Fig. 10. The enhancement image of different algorithms

Download full size | View in the Article

Fig. 11. Comparison of image enhancement effects by the method in this paper

Download full size | View in the Article

Kernels	Scales	En	EPI	Time
3	0.1, 0.5, 1	5.63	4.20	0.28
6	0.1, 0.3, 0.5, 0.7, 0.9, 1	6.39	4.96	0.53
9	0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.9, 1	6.41	5.01	0.81

Table 1. Running time of MSC with different parameters and En、EPI of images processed by MSC

View in the Article

$γ$	AG	EME	En
0	3.435 3	4.369 8	5.360 3
0.1	3.436 8	4.370 0	5.444 5
0.2	3.444 0	4.421 2	5.598 6
0.3	3.685 0	4.632 8	5.685 6
0.4	4.232 0	5.002 1	6.002 3
0.5	4.498 5	5.332 3	6.112 3
0.6	4.902 1	5.952 0	6.399 8
0.7	5.253 6	6.232 9	6.860 0
0.8	6.360 2	7.033 4	7.368 8
0.9	4.553 6	5.369 8	6.998 5
1	4.033 2	5.332 0	6.523 0

Table 2. AG, EME, and En of enhanced images by BABHE with different

γ

View in the Article

Image	Evaluation index	Original	BBHE	HEEF	DOTHE	RLBHE	FCCE	Wan	Our
Image A	EME	5.219 7	6.959 3	12.022 9	8.069 1	3.893 2	7.998 5	2.144 6	15.139 1
	En	7.071 1	7.293 5	6.773 2	7.639 2	6.526 1	7.117 2	6.839 9	7.846 8
	AG	3.586 3	5.154 7	6.296 3	6.398 6	2.531 5	3.999 7	3.685 3	13.718 8
	t	-	0.632 5	0.017 8	11.176 6	0.867 2	0.423 0	58.029 5	16.536 5
Image B	EME	1.457 2	8.454 9	3.477 8	9.195 0	2.856 0	7.222 9	0.951 0	13.466 6
	En	5.820 8	7.642 5	5.690 6	7.655 7	6.614 2	7.535 8	5.857 8	7.525 4
	AG	2.415 9	9.283 0	4.529 0	8.876 2	4.833 9	8.196 0	2.467 7	15.863 7
	t	-	0.510 3	0.012 9	12.364 4	0.908 0	0.392 1	56.983 3	16.145 2
Image C	EME	1.832 2	6.415 3	2.024 7	5.273 4	1.816 7	1.655 3	1.133 2	7.067 9
	En	5.039 1	7.523 9	3.857 2	7.856 0	6.144 5	5.768 3	5.533 6	7.163 3
	AG	1.142 6	6.099 0	0.740 0	7.864 3	2.433 2	5.363 9	1.337 5	11.988 4
	t	-	0.600 2	0.150 4	16.010 4	1.017 2	0.446 8	52.362 3	16.688 0
Image D	EME	3.178 9	8.685 8	9.287 2	7.347 4	3.094 6	2.095 5	0.941 0	17.322 8
	En	5.468 1	6.895 0	5.317 3	7.634 1	5.589 2	5.874 1	5.409 6	7.290 4
	AG	1.782 0	5.341 9	4.156 7	5.709 2	1.243 6	2.104 9	2.203 5	13.991 8
	t	-	0.663 5	0.018 5	14.450 0	1.106 0	0.432 0	61.036 5	14.064 7
Image E	EME	0.821 4	7.209 9	2.063 2	6.452 3	2.391 0	4.575 7	0.767 1	12.594 9
	En	5.425 6	6.766 1	5.250 1	7.762 5	6.617 1	7.650 5	7.310 3	7.809 1
	AG	1.001 5	3.088 8	2.021 9	6.523 1	2.442 5	5.435 4	4.785 2	11.516 9
	t	-	0.234 5	0.011 2	9.972 4	0.869 3	0.274 0	58.258 0	10.694 6
Image F	EME	13.446 1	23.586 9	75.901 7	32.311 5	23.540 1	26.066 4	4.360 1	90.252 5
	En	6.587 4	7.014 2	6.321 8	7.306 8	6.409 3	7.403 0	6.331 5	7.540 3
	AG	7.845 0	9.748 9	22.313 3	19.900 7	11.317 3	14.236 5	6.889 9	53.617 7
	t	-	0.223 5	0.006 6	7.011 6	0.678 6	0.139 8	22.406 9	6.303 9

Table 4. Running time of different algorithms and EME, En and AG of processed image by different algorithms

View in the Article

	Original image	HEEF	BBHE	BABHE
Average grays	57	46	68	88
Number of gray levels	232	254	256	256
En	6.635 7	6.554 3	7.272 4	7.368 8

Table 4. Objective evaluation of image quality after enhancement processing by each algorithm

View in the Article

Input	MSC/s	BABHE/s	ALL/s	LF/s
Image A	0.615 9	15.485 4	0.058 7	0.376 5
Image B	0.577 8	15.504 4	0.060 3	0.001 3
Image C	0.553 3	16.032 5	0.068 5	0.033 6
Image D	0.602 3	13.336 2	0.060 3	0.065 9
Image E	0.583 8	10.048 2	0.061 7	0.000 9
Image F	0.360 3	5.881 6	0.061 4	0.000 6

Table 5. Running time of each stage of the algorithm in this paper

Hao-xiang LU, Zhen-bing LIU, Peng-yue GUO, Xi-peng PAN. Multi-scale Convolution Combined with Adaptive Bi-interval Equalization for Image Enhancement[J]. Acta Photonica Sinica, 2020, 49(10): 1010002

Download Citation

Tools

Save the article for my favorites

Paper Information