图像重建是光学计算成像的关键环节之一。目前基于深度学习的图像重建主要使用卷积神经网络、循环神经网络或生成对抗网络等模型。大多数研究仅通过单一模态的数据训练模型,难以在保证成像质量的同时又具备不同场景的泛化能力。为解决...图像重建是光学计算成像的关键环节之一。目前基于深度学习的图像重建主要使用卷积神经网络、循环神经网络或生成对抗网络等模型。大多数研究仅通过单一模态的数据训练模型,难以在保证成像质量的同时又具备不同场景的泛化能力。为解决这一问题,提出了一种基于Transformer模块的多模态图像重建模型(multi-modal image reconstruction model based on the Transformer,Trans-MIR)。实验结果表明,Trans-MIR能够从多模态数据中提取图像特征,实现高质量的图像重建,对二维通用人脸散斑图像进行图像重建的结构相似度高达0.93,对三维微管结构图像的超分辨重建的均方误差低至10^(−4)量级。Trans-MIR对研究多模态图像重建具有一定的启发作用。展开更多
In this paper, we investigate the weighted iterative decoding to improve the performance of turbo-polar code. First of all, a minimum weighted mean square error criterion is proposed to optimize the scaling factors(SF...In this paper, we investigate the weighted iterative decoding to improve the performance of turbo-polar code. First of all, a minimum weighted mean square error criterion is proposed to optimize the scaling factors(SFs). Secondly, for two typical iterative algorithms,such as soft cancellation(SCAN) and belief propagation(BP) decoding, genie-aided decoders are proposed as the ideal reference of the practical decoding. Guided by this optimization framework, the optimal SFs of SCAN or BP decoders are obtained. The bit error rate performance of turbo-polar code with the optimal SFs can achieve 0.3 dB or 0.7 dB performance gains over the standard SCAN or BP decoding respectively.展开更多
Edge detection for low-contrast phase objects cannot be performed directly by the spatial difference of intensity distribution.In this work,an all-optical diffractive neural network(DPENet)based on the differential in...Edge detection for low-contrast phase objects cannot be performed directly by the spatial difference of intensity distribution.In this work,an all-optical diffractive neural network(DPENet)based on the differential interference contrast principle to detect the edges of phase objects in an all-optical manner is proposed.Edge information is encoded into an interference light field by dual Wollaston prisms without lenses and light-speed processed by the diffractive neural network to obtain the scale-adjustable edges.Simulation results show that DPENet achieves F-scores of 0.9308(MNIST)and 0.9352(NIST)and enables real-time edge detection of biological cells,achieving an F-score of 0.7462.展开更多
文摘图像重建是光学计算成像的关键环节之一。目前基于深度学习的图像重建主要使用卷积神经网络、循环神经网络或生成对抗网络等模型。大多数研究仅通过单一模态的数据训练模型,难以在保证成像质量的同时又具备不同场景的泛化能力。为解决这一问题,提出了一种基于Transformer模块的多模态图像重建模型(multi-modal image reconstruction model based on the Transformer,Trans-MIR)。实验结果表明,Trans-MIR能够从多模态数据中提取图像特征,实现高质量的图像重建,对二维通用人脸散斑图像进行图像重建的结构相似度高达0.93,对三维微管结构图像的超分辨重建的均方误差低至10^(−4)量级。Trans-MIR对研究多模态图像重建具有一定的启发作用。
基金supported by the National Natural Science Foundation of China(No.61671080)the National Natural Science Foundation of China(No.61771066)Nokia Beijing Bell Lab
文摘In this paper, we investigate the weighted iterative decoding to improve the performance of turbo-polar code. First of all, a minimum weighted mean square error criterion is proposed to optimize the scaling factors(SFs). Secondly, for two typical iterative algorithms,such as soft cancellation(SCAN) and belief propagation(BP) decoding, genie-aided decoders are proposed as the ideal reference of the practical decoding. Guided by this optimization framework, the optimal SFs of SCAN or BP decoders are obtained. The bit error rate performance of turbo-polar code with the optimal SFs can achieve 0.3 dB or 0.7 dB performance gains over the standard SCAN or BP decoding respectively.
基金supported by the National Key Research and Development Program of China(Nos.2021YFB2802000 and 2022YFB2804301)Shanghai Municipal Science and Technology Major Project,Science and Technology Commission of Shanghai Municipality(No.21DZ1100500)+2 种基金Shanghai Frontiers Science Center Program(2021-2025 No.20)National Natural Science Foundation of China(Nos.61975123 and 12072200)Science and Technology Development Foundation of Pudong New Area(No.PKX2021-D10)。
文摘Edge detection for low-contrast phase objects cannot be performed directly by the spatial difference of intensity distribution.In this work,an all-optical diffractive neural network(DPENet)based on the differential interference contrast principle to detect the edges of phase objects in an all-optical manner is proposed.Edge information is encoded into an interference light field by dual Wollaston prisms without lenses and light-speed processed by the diffractive neural network to obtain the scale-adjustable edges.Simulation results show that DPENet achieves F-scores of 0.9308(MNIST)and 0.9352(NIST)and enables real-time edge detection of biological cells,achieving an F-score of 0.7462.
