Although Convolutional Neural Networks(CNNs)have significantly improved the development of image Super-Resolution(SR)technology in recent years,the existing SR methods for SAR image with large scale factors have rarel...Although Convolutional Neural Networks(CNNs)have significantly improved the development of image Super-Resolution(SR)technology in recent years,the existing SR methods for SAR image with large scale factors have rarely been studied due to technical difficulty.A more efficient method is to obtain comprehensive information to guide the SAR image reconstruction.Indeed,the co-registered High-Resolution(HR)optical image has been successfully applied to enhance the quality of SAR image due to its discriminative characteristics.Inspired by this,we propose a novel Optical-Guided Super-Resolution Network(OGSRN)for SAR image with large scale factors.Specifically,our proposed OGSRN consists of two sub-nets:a SAR image SuperResolution U-Net(SRUN)and a SAR-to-Optical Residual Translation Network(SORTN).The whole process during training includes two stages.In stage-1,the SR SAR images are reconstructed by the SRUN.And an Enhanced Residual Attention Module(ERAM),which is comprised of the Channel Attention(CA)and Spatial Attention(SA)mechanisms,is constructed to boost the representation ability of the network.In stage-2,the output of the stage-1 and its corresponding HR SAR images are translated to optical images by the SORTN,respectively.And then the differences between SR images and HR images are computed in the optical space to obtain feedback information that can reduce the space of possible SR solution.After that,we can use the optimized SRUN to directly produce HR SAR image from Low-Resolution(LR)SAR image in the testing phase.The experimental results show that under the guidance of optical image,our OGSRN can achieve excellent performance in both quantitative assessment metrics and visual quality.展开更多
The number of bamboo stem at different ages and the mean diameter at breast height(DBH)which are the important target in evaluating productivity of bamboo stand were investigated in 50 plots established in Jianou city...The number of bamboo stem at different ages and the mean diameter at breast height(DBH)which are the important target in evaluating productivity of bamboo stand were investigated in 50 plots established in Jianou city, Fujian Province in this paper, and the authors selected the method of artificial neural network to biuld the simulative and predictive model of mean DBH for bamboo stands. Artificial neural network is a good method in handling the overall nonlinear mapping problems between input variables and output ones, which has a wide application in many research fields, such as system simulating, automation controlling, paralleled data processing and so on. In this paper, the input variables were the number of different age and the total number of stand, the output variable was mean DBH for bamboo stands, the number of neurons of hide level( M ) was M=2L+1=3 according to the last document ( L is the number of factors of input level), and the network activity function is Sigmiod function as follows: F(x)=1/(1+e -x ). Using the built BP network, the samples were trained until E j(W 1 lm ,W 2 mn )=Nn=1(O nj -Y nj ) 2 =min, where O nj and Y nj are output values of network and really values of DBH for bamboo stands respectively, N is the number of trained samples, and E j is sum of square deviation of BP network. If E j didn’t converge, the weights and thresholds of BP network were adjusted as follow: ΔW ij (n+1)=βλ jX i+αΔW ij (n) and Δη j(n+1)=-βλ j+αΔη j(n) .. The results showed that the mean simulative accuracy and the mean predictive accuracy of mean D.B.H BP model for bamboo stands were all satisfactory, which were 89 95% and 89 26% respectively. Therefore, it provided a scientific basis for evaluating the productivity and realizing high yield for bamboo stands.展开更多
基金supported by the National Natural Science Foundation of China(Nos.61771319,62076165 and 61871154)the Natural Science Foundation of Guangdong Province,China(No.2019A1515011307)+1 种基金Shenzhen Science and Technology Project,China(Nos.JCYJ20180507182259896 and 20200826154022001)the other project(Nos.2020KCXTD004 and WDZC20195500201)。
文摘Although Convolutional Neural Networks(CNNs)have significantly improved the development of image Super-Resolution(SR)technology in recent years,the existing SR methods for SAR image with large scale factors have rarely been studied due to technical difficulty.A more efficient method is to obtain comprehensive information to guide the SAR image reconstruction.Indeed,the co-registered High-Resolution(HR)optical image has been successfully applied to enhance the quality of SAR image due to its discriminative characteristics.Inspired by this,we propose a novel Optical-Guided Super-Resolution Network(OGSRN)for SAR image with large scale factors.Specifically,our proposed OGSRN consists of two sub-nets:a SAR image SuperResolution U-Net(SRUN)and a SAR-to-Optical Residual Translation Network(SORTN).The whole process during training includes two stages.In stage-1,the SR SAR images are reconstructed by the SRUN.And an Enhanced Residual Attention Module(ERAM),which is comprised of the Channel Attention(CA)and Spatial Attention(SA)mechanisms,is constructed to boost the representation ability of the network.In stage-2,the output of the stage-1 and its corresponding HR SAR images are translated to optical images by the SORTN,respectively.And then the differences between SR images and HR images are computed in the optical space to obtain feedback information that can reduce the space of possible SR solution.After that,we can use the optimized SRUN to directly produce HR SAR image from Low-Resolution(LR)SAR image in the testing phase.The experimental results show that under the guidance of optical image,our OGSRN can achieve excellent performance in both quantitative assessment metrics and visual quality.
文摘The number of bamboo stem at different ages and the mean diameter at breast height(DBH)which are the important target in evaluating productivity of bamboo stand were investigated in 50 plots established in Jianou city, Fujian Province in this paper, and the authors selected the method of artificial neural network to biuld the simulative and predictive model of mean DBH for bamboo stands. Artificial neural network is a good method in handling the overall nonlinear mapping problems between input variables and output ones, which has a wide application in many research fields, such as system simulating, automation controlling, paralleled data processing and so on. In this paper, the input variables were the number of different age and the total number of stand, the output variable was mean DBH for bamboo stands, the number of neurons of hide level( M ) was M=2L+1=3 according to the last document ( L is the number of factors of input level), and the network activity function is Sigmiod function as follows: F(x)=1/(1+e -x ). Using the built BP network, the samples were trained until E j(W 1 lm ,W 2 mn )=Nn=1(O nj -Y nj ) 2 =min, where O nj and Y nj are output values of network and really values of DBH for bamboo stands respectively, N is the number of trained samples, and E j is sum of square deviation of BP network. If E j didn’t converge, the weights and thresholds of BP network were adjusted as follow: ΔW ij (n+1)=βλ jX i+αΔW ij (n) and Δη j(n+1)=-βλ j+αΔη j(n) .. The results showed that the mean simulative accuracy and the mean predictive accuracy of mean D.B.H BP model for bamboo stands were all satisfactory, which were 89 95% and 89 26% respectively. Therefore, it provided a scientific basis for evaluating the productivity and realizing high yield for bamboo stands.
