Infrared Image Fusion Algorithm Selection Based on Joint Drop Shadow of Possibility Distributions

Linna JI; Xiaoming GUO; Fengbao YANG; Yaling ZHANG

doi:10.3788/gzxb20215004.0410001

Journals >Acta Photonica Sinica >Volume 50 >Issue 4 >Page 228 > Article

Acta Photonica Sinica
Vol. 50, Issue 4, 228 (2021)

Infrared Image Fusion Algorithm Selection Based on Joint Drop Shadow of Possibility Distributions

Linna JI, Xiaoming GUO, Fengbao YANG, and Yaling ZHANG

Author Affiliations

School of Information and Communication Engineering， North University of China， Taiyuan030051， China

show less

DOI: 10.3788/gzxb20215004.0410001 Cite this Article

Linna JI, Xiaoming GUO, Fengbao YANG, Yaling ZHANG. Infrared Image Fusion Algorithm Selection Based on Joint Drop Shadow of Possibility Distributions[J]. Acta Photonica Sinica, 2021, 50(4): 228 Copy Citation Text

show less

Fig. 1. Flow chart of dual-mode infrared image fusion algorithm selection

Download full size | View in the Article

Fig. 2. Source dual-mode infrared images of two groups

Download full size | View in the Article

Fig. 3. Fusion image results based on fusion algorithm sets

Download full size | View in the Article

Fig. 4. Difference feature amplitude intensity distribution

Download full size | View in the Article

Fig. 5. Fusion effective degree distribution of difference feature amplitude

Download full size | View in the Article

Fig. 6. Probability density distribution of difference feature amplitude

Download full size | View in the Article

Fig. 7. Heterogeneous difference feature weight function fusion distribution synthesis

Download full size | View in the Article

Fig. 8. Heterogeneous difference feature weight function fusion distribution joint drop shadow

Download full size | View in the Article

Fig. 9. Evaluation index of different fusion algorithms

Download full size | View in the Article

Fig. 10. R_i of different fusion algorithms

Download full size | View in the Article

T-module operator	Correlation of x, y	$p_{x, y}$
$T_{1} (x, y) = m a x (0, x + y - 1)$	Extremely negative correlation	$[- 1, - 0.5)$
$T_{2} (x, y) = m a x (0, (x^{0.5} + y^{0.5} {- 1))}^{2}$	Negative correlation	$[- 0.5,0)$
$T_{3} (x, y) = x \cdot y$	Irrelevant	$[0,0.2)$
$T_{4} (x, y) = (x^{- 0.5} + y^{- 0.5} {C 1)}^{- 0.5}$	Weak positive correlation	$[0.2,0.4)$
$T_{5} (x, y) = (x^{- 1} + y^{- 1} {- 1)}^{- 1}$	Positive correlation	$[0.4,0.6)$
$T_{6} (x, y) = m i n (x, y)$	Extremely positive correlation	$[0.6,1)$

Table 1. Operational rules of T-module operator

View in the Article

Feature weight function	Q₁、Q₂	Q₂、Q₃	Q₃、Q₄	Q₄、Q₁	Q₂、Q₄	Q₁、Q₃
T-module operator	T₃	T₃	T₃	T₆	T₆	T₆

Table 2. The first group of heterogeneous difference feature weight function synthesis rules

View in the Article

Feature weight function	Q₁、Q₂	Q₂、Q₃	Q₃、Q₄	Q₄、Q₁	Q₂、Q₄	Q₁、Q₃
T-module operator	T₄	T₄	T₄	T₆	T₆	T₆

Table 3. The second group of heterogeneous difference feature weight function synthesis rules

View in the Article

Difference features	Index	Fusion algorithm
Difference features	Index	A₁	A₂	A₃	A₄	A₅	A₆	A₇	A₈	A₉	A₁₀	A₁₁
Q₁、Q₂	$f_{i}$	0.014 8	0	0	0.023 7	0.005 9	0.020 7	0.221 9	0.665 7	0	0.005 9	0.041 4
	$\bar{E_{i}}$	0.222 6	0	0	0.381 3	0.598 0	0.214 0	0.210 1	0.343 9	0	0.203 1	0.266 5
	D_i	0.003 3	0	0	0.009 0	0.0035	0.004 4	0.046 6	0.228 9	0	0.001 2	0.011 0
Q₁、Q₃	$f_{i}$	0.009 4	0	0.128 8	0.025 8	0.264 6	0.103 0	0.058 5	0.405 2	0	0	0.004 7
	$\bar{E_{i}}$	0.268 8	0	0.426 7	0.312 5	0.333 3	0.334 1	0.143 0	0.311 8	0	0	0.228 2
	D_i	0.002 5	0	0.055 0	0.008 1	0.088 2	0.034 4	0.008 4	0.126 3	0	0	0.001 1
Q₁、Q₄	$f_{i}$	0.069 7	0.014 9	0.064 7	0.119 4	0	0.159 2	0.174 1	0.338 3	0.010 0	0	0.049 8
	$\bar{E_{i}}$	0.291 8	0.281 4	0.269 1	0.539 3	0	0.181 8	0.191 7	0.356 1	0.226 4	0	0.205 6
	D_i	0.020 3	0.004 2	0.017 4	0.064 4	0	0.028 9	0.033 4	0.120 5	0.002 3	0	0.010 2
Q₂、Q₃	$f_{i}$	0.022 6	0	0.057 8	0.085 4	0.030 2	0.163 3	0.198 5	0.429 6	0	0	0.012 6
	$\bar{E_{i}}$	0.655 0	0	0.398 1	0.565 0	0.540 0	0.372 7	0.212 0	0.500 5	0	0	0.665 1
	D_i	0.014 8	0	0.023 0	0.048 3	0.016 3	0.060 9	0.042 1	0.215 1	0	0	0.008 4
Q₂、Q₄	$f_{i}$	0.042 7	0	0	0.042 7	0.273 9	0.211 1	0.123 1	0.256 3	0	0	0.050 3
	$\bar{E_{i}}$	0.659 9	0	0	0.575 5	0.448 0	0.244 0	0.255 3	0.489 2	0	0	0.722 6
	D_i	0.028 2	0	0	0.024 6	0.122 7	0.051 5	0.031 4	0.125 4	0	0	0.036 3
Q₃、Q₄	$f_{i}$	0.025 3	0	0.098 5	0.093 4	0.313 1	0.260 1	0.101 0	0.108 6	0	0	0
	$\bar{E_{i}}$	0.747 1	0	0.321 3	0.600 9	0.451 0	0.321 0	0.189 4	0.595 1	0	0	0
	D_i	0.018 9	0	0.031 6	0.056 1	0.141 2	0.083 5	0.019 1	0.064 6	0	0	0
Sum		0.088 0	0.004 2	0.127 0	0.210 5	0.371 9	0.263 6	0.181 0	0.880 8	0.002 3	0.001 2	0.067 0
Rank		7	9	6	4	2	3	5	1	10	11	8

Table 4. D_iof different fusion algorithms of the first group

View in the Article

Difference features	Index	Fusion algorithm
Difference features	Index	A₁	A₂	A₃	A₄	A₅	A₆	A₇	A₈	A₉	A₁₀	A₁₁
Q₁、Q₂	$f_{i}$	0	0.115 8	0	0.126 3	0	0	0.255 3	0.047 4	0.018 4	0.328 9	0.107 9
	$\bar{E_{i}}$	0	0.687 6	0	0.780 5	0	0	0.689 1	0.812 4	0.849 2	0.636 1	0.719 8
	D_i	0	0.079 6	0	0.098 6	0	0	0.175 9	0.038 5	0.015 6	0.209 2	0.077 7
Q₁、Q₃	$f_{i}$	0.007 2	0.007 2	0	0	0.033 8	0.079 7	0.108 7	0.152 2	0	0.596 6	0.014 5
	$\bar{E_{i}}$	0.675 5	0.771 7	0	0	0.704 8	0.797 5	0.697 4	0.785 8	0	0.679 0	0.654 9
	D_i	0.004 9	0.005 6	0	0	0.023 8	0.063 6	0.075 8	0.119 6	0	0.405 1	0.009 5
Q₁、Q₄	$f_{i}$	0	0.257 2	0	0.101 2	0	0	0.274 6	0.106 9	0.040 5	0.219 7	0
	$\bar{E_{i}}$	0	0.696 9	0	0.774 6	0	0	0.685 7	0.824 9	0.645 6	0.571 4	0
	D_i	0	0.179 3	0	0.078 4	0	0	0.188 3	0.088 2	0.026 1	0.125 5	0
Q₂、Q₃	$f_{i}$	0	0	0	0.042 5	0	0.015 0	0	0.192 5	0	0.667 5	0.082 5
	$\bar{E_{i}}$	0	0	0	0.414 6	0	0.505 6	0	0.491 6	0	0.265 2	0.283 9
	D_i	0	0	0	0.017 6	0	0.007 6	0	0.094 6	0	0.177 0	0.023 4
Q₂、Q₄	$f_{i}$	0	0.145 7	0.002 5	0.138 2	0.055 3	0	0	0.077 9	0	0.434 7	0.145 7
	$\bar{E_{i}}$	0	0.285 0	0.307 4	0.452 5	0.346 5	0	0	0.484 4	0	0.166 8	0.401 5
	D_i	0	0.041 5	0.000 8	0.062 5	0.019 2	0	0	0.037 7	0	0.072 5	0.058 5
Q₃、Q₄	$f_{i}$	0	0.072 1	0.004 5	0.038 3	0.018 0	0.009 0	0	0.254 5	0	0.601 4	0.002 3
	$\bar{E_{i}}$	0	0.323 6	0.261 1	0.379 9	0.342 6	0.738 2	0	0.432 1	0	0.199 5	0.610 5
	D_i	0	0.023 3	0.001 2	0.014 5	0.006 2	0.006 7	0	0.110 0	0	0.120 0	0.001 4
Sum		0.004 9	0.329 3	0.001 9	0.271 6	0.049 2	0.077 8	0.440 0	0.488 6	0.041 8	1.109 4	0.170 5
Rank		10	4	11	5	8	7	3	2	9	1	6

Table 5. D_iof different fusion algorithms of the second group

View in the Article

Image	Method	Evaluation index								R_i	Rank
Image	Method	X₁	X₂	X₃	X₄	X₅	X₆	X₇	X₈	R_i	Rank
1	A₁	0.534 2	0.454 7	0.563 1	0.224 1	0.186 4	0.447 7	0.354 4	0.066 9	37	3
	A₂	0.391 2	0.394 0	0.612 7	0.136 2	0.252 9	0.387 3	0.242 1	0.057 1	70	10
	A₃	0.419 0	0.434 3	0.613 9	0.181 6	0.473 0	0.416 0	0.364 5	0.081 5	45	5
	A₄	0.541 5	0.435 4	0.634 4	0.247 7	0.462 7	0.495 5	0.693 9	0.067 2	30	2
	A₅	0.410 7	0.434 1	0.625 2	0.173 0	0.471 6	0.412 6	0.364 2	0.079 1	50	7
	A₆	0.560 3	0.495 0	0.473 0	0.222 9	0.330 2	0.495 1	0.835 1	0.057 6	39	4
	A₇	0.045 9	0.001 4	0.003 5	0.000 1	0.004 4	0.000 7	0.100 4	0.060 9	77	11
	A₈	0.7028	0.485 8	0.669 0	0.222 0	0.477 0	0.449 8	0.776 6	0.082 3	20	1
	A₉	0.452 5	0.429 1	0.648 2	0.158 4	0.383 3	0.418 0	0.275 4	0.083 3	48	6
	A₁₀	0.360 0	0.437 9	0.617 1	0.134 6	0.391 2	0.428 4	0.294 8	0.061 4	58	9
	A₁₁	0.421 5	0.398 8	0.660 0	0.145 4	0.383 5	0.390 9	0.252 5	0.082 7	54	8
2	A₁	0.455 5	0.466 7	0.715 8	0.261 0	0.454 2	0.452 2	0.418 1	0.020 6	46	5
	A₂	0.542 4	0.396 1	0.770 5	0.294 4	0.438 3	0.384 6	0.275 4	0.030 7	50	7
	A₃	0.468 4	0.423 2	0.673 3	0.265 7	0.442 8	0.393 7	0.420 1	0.033 1	51	8
	A₄	0.535 2	0.502 6	0.539 1	0.281 5	0.442 0	0.494 6	0.709 9	0.035 3	37	3
	A₅	0.403 0	0.402 0	0.618 2	0.239 5	0.413 2	0.370 2	0.418 7	0.030 5	68	11
	A₆	0.524 2	0.523 7	0.537 0	0.249 5	0.343 6	0.516 2	0.777 4	0.020 2	51	9
	A₇	0.522 7	0.420 8	0.748 3	0.286 0	0.567 9	0.402 2	0.301 5	0.028 3	45	4
	A₈	0.502 9	0.402 0	0.734 0	0.233 6	0.409 9	0.382 9	0.350 6	0.025 6	67	10
	A₉	0.545 4	0.430 1	0.780 0	0.285 6	0.522 0	0.410 5	0.309 5	0.034 3	35	2
	A₁₀	0.571 5	0.527 3	0.536 3	0.295 6	0.549 9	0.493 1	0.739 5	0.024 8	31	1
	A₁₁	0.524 7	0.388 9	0.785 8	0.285 7	0.548 7	0.380 6	0.272 7	0.036 2	47	6

Table 6. Evaluation index of different fusion algorithms of images

Linna JI, Xiaoming GUO, Fengbao YANG, Yaling ZHANG. Infrared Image Fusion Algorithm Selection Based on Joint Drop Shadow of Possibility Distributions[J]. Acta Photonica Sinica, 2021, 50(4): 228

Download Citation

Tools

Save the article for my favorites

Paper Information