Laser & Optoelectronics Progress, Volume. 60, Issue 10, 1028004(2023)
Non-Subsampling Shearlet Transform Remote Sensing Image Fusion with Improved Dual-channel Adaptive Pulse Coupled Neural Network
Figures & Tables(17)
Fig. 1. Frequency domain subdivision diagram and support interval of NSST. (a) Frequency domain subdivision map; (b) frequency domain support interval
Fig. 3. Fusion results of high-frequency coefficients in all directions. (a)-(b) High frequency subbands in 2 directions in first layer; (c)-(f) high frequency subbands in 4 directions in second layer
Fig. 8. Fusion results of first group. (a) SE; (b) NSCT; (c) ISCM; (d) WDCPAPCNN; (e) PAPCNN; (f) proposed method
Fig. 10. Fusion results of second group. (a) SE; (b) NSCT; (c) ISCM; (d) WDCPAPCNN; (e) PAPCNN; (f) proposed method
Fig. 12. Fusion results of third group. (a) SE; (b) NSCT; (c) ISCM; (d) WDCPAPCNN; (e) PAPCNN; (f) proposed method
Table 1. Quantitative evaluation of spatial information in all directions
View tableTable 1. Quantitative evaluation of spatial information in all directions
Index Method 1 2 3 4 5 6 AG DCPCNN 0.0087 0.0072 0.0043 0.0033 0.0036 0.0044 IDCPCNN 0.0067 0.0056 0.0017 0.0023 0.0016 0.0024 SF DCPCNN 0.0235 0.0190 0.0149 0.0115 0.0123 0.0138 IDCPCNN 0.0216 0.0173 0.0095 0.0098 0.0079 0.0102 STD DCPCNN 0.0211 0.0163 0.0094 0.0071 0.0074 0.0084 IDCPCNN 0.0181 0.0140 0.0064 0.0060 0.0051 0.0061
Table 2. Settings of NSST decomposition layers and corresponding directions
View tableTable 2. Settings of NSST decomposition layers and corresponding directions
Decomposition layers Direction number 1 16 2 16，16 3 16，16，8 4 16，16，8，8 5 16，16，8，8，4
Table 3. Quantitative evaluation of first group of experiments
View tableTable 3. Quantitative evaluation of first group of experiments
Image Method AG↑ SF↑ QAB/F↑ VIFF↑ E↑ FMI↑ 1 SE 0.0068 0.0197 0.3127 0.5634 4.4744 0.9078 NSCT 0.0163 0.0407 0.7008 0.9523 6.8287 0.8894 ISCM 0.0138 0.0370 0.6261 0.8966 6.9263 0.9122 WDCPA-PCNN 0.0126 0.0313 0.5040 0.8457 6.9216 0.8986 PAPCNN 0.0169 0.0426 0.6050 0.9462 6.9746 0.9034 Proposed method 0.0175 0.0450 0.6353 0.9901 6.9823 0.9126 2 SE 0.0078 0.0152 0.4083 0.5834 6.4831 0.8717 NSCT 0.0143 0.0381 0.6802 0.8630 6.3622 0.8555 ISCM 0.0126 0.0325 0.6113 0.8421 6.5496 0.8685 WDCPA-PCNN 0.0124 0.0306 0.5178 0.8435 6.6625 0.8534 PAPCNN 0.0156 0.0393 0.5800 0.8946 6.5852 0.8394 Proposed method 0.0166 0.0447 0.6648 0.8993 6.5802 0.8744
Table 4. Quantitative evaluation of second group of experiments
View tableTable 4. Quantitative evaluation of second group of experiments
Image Method AG↑ SF↑ QAB/F↑ VIFF↑ E↑ FMI↑ 1 SE 0.0065 0.0153 0.3803 0.6260 5.5879 0.9325 NSCT 0.0082 0.0213 0.5522 0.4642 6.0160 0.9117 ISCM 0.0069 0.0180 0.3918 0.5078 6.1748 0.9234 WDCPA-PCNN 0.0070 0.0183 0.3630 0.5041 6.1924 0.9163 PAPCNN 0.0087 0.0222 0.5281 0.5495 6.1991 0.9122 Proposed method 0.0090 0.0240 0.6087 0.5689 6.1754 0.9250 2 SE 0.0051 0.0113 0.4257 0.6167 6.0126 0.8975 NSCT 0.0137 0.0364 0.5215 1.0716 6.4405 0.8705 ISCM 0.0113 0.0301 0.4313 1.0154 6.5779 0.8877 WDCPA-PCNN 0.0125 0.0324 0.4954 1.0205 6.6032 0.8732 PAPCNN 0.0155 0.0410 0.5107 1.0150 6.6462 0.8523 Proposed method 0.0158 0.0426 0.4955 1.0500 6.6329 0.8886
Table 5. Quantitative evaluation of third group of experiments
View tableTable 5. Quantitative evaluation of third group of experiments
Image Method AG↑ SF↑ QAB/F↑ VIFF↑ E↑ FMI↑ 1 SE 0.0033 0.0084 0.5525 0.7099 5.8743 0.9318 NSCT 0.0049 0.0157 0.6477 0.9820 5.8277 0.9404 ISCM 0.0045 0.0152 0.6562 0.9748 6.0193 0.9447 WDCPA-PCNN 0.0045 0.0124 0.5241 0.8857 6.0950 0.9248 PAPCNN 0.0051 0.0143 0.5814 0.9827 6.0443 0.9318 Proposed method 0.0053 0.0164 0.6406 0.9898 6.0316 0.9447 2 SE 0.0047 0.0093 0.5302 0.7573 5.7709 0.9059 NSCT 0.0051 0.0123 0.6789 0.6124 5.3645 0.8900 ISCM 0.0049 0.0110 0.6191 0.6921 5.5513 0.9063 WDCPA-PCNN 0.0058 0.0141 0.5905 0.7971 5.6667 0.8825 PAPCNN 0.0057 0.0132 0.6054 0.7394 5.5513 0.8936 Proposed method 0.0061 0.0146 0.6693 0.7474 5.5751 0.9061
Tools
Get Citation
Copy Citation Text
Linian Ruan, Yan Dong. Non-Subsampling Shearlet Transform Remote Sensing Image Fusion with Improved Dual-channel Adaptive Pulse Coupled Neural Network[J]. Laser & Optoelectronics Progress, 2023, 60(10): 1028004
Paper Information
Category: Remote Sensing and Sensors
Received: Nov. 3, 2021
Accepted: Feb. 14, 2022
Published Online: May. 17, 2023
The Author Email: Dong Yan (dongyanchina@sina.com)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20