红外与激光工程
2022, 51(2): 20210902
南京理工大学 电子工程与光电技术学院，江苏 南京 210094
提出了一种红外热成像视频中运动目标识别追踪以及轨迹重构的动态仿真方法。通过仿真环境中虚拟红外图像的生成方式与成像的基本模型，对得到的图像进行一系列预处理。以空对空场景搭建了基于Gazebo与OpenCV的动态仿真平台，利用平滑约束算法对追踪目标进行实时动态轨迹重构，提出了误差分析模型，并分析出轨迹重构算法的性能以及仿真平台的效能。实验结果表明，该方法针对空对空场景下的红外运动目标轨迹重构具有较好的精度与鲁棒性，对目标的运动模型基本没有约束，同时仿真平台具有较高的运行效能与实时性，普通家用电脑即可实现高于60 fps的实时动态仿真，满足轨迹重构算法性能测试与训练的需求，其核心算法亦可迁移至机载计算平台实现真实场景下的实时轨迹重构。所提出的单路热成像视频中运动目标轨迹重构动态仿真方法对空间目标三维轨迹重建与动态测距定位的研究具有重要意义。
动态仿真 红外成像 轨迹重构 平滑约束 dynamic simulation infrared imaging trajectory reconstruction smoothing constraint 红外与激光工程
2022, 51(2): 20210901
南京理工大学 电子工程与光电技术学院，江苏 南京 210094
卷积作为一种简单的线性平移不变运算，被广泛应用于图像处理的各个领域，其衍生出的卷积神经网络更是在人工智能领域中大放异彩。为了应对后摩尔时代AI推理芯片算力受限的问题，光学神经网络应运而生。光学卷积神经网络作为其中一个重要的研究热点对光学神经网络的发展起到了重要的推动作用。设计了一种光学卷积系统，基于微透镜阵列与透镜组成的匀光光路对光场所携带的图像做二维卷积，该系统可以光学实现图像平滑和锐化。当使用空间光调制器来投影卷积核和输入图像时，系统可以实现各种步长的三种卷积形式，也可以通过多次投影/平铺实现多通道的三维卷积，进而为实现光学卷积神经网络用于复杂的图像处理任务奠定基础。
光学卷积 微透镜阵列 匀光系统 图像处理 optical convolution microlens array unifying system image processing 红外与激光工程
2022, 51(2): 20210887
Author Affiliations
Abstract
1 School of Electronic and Optical Engineering, Nanjing University of Science and Technology, Nanjing 210094, China
2 Institute of Armored Forces, Army Research Institute, Beijing, China
With the advent of the era of big data, artificial intelligence has attracted continuous attention from all walks of life, and has been widely used in medical image analysis, molecular and material science, language recognition and other fields. As the basis of artificial intelligence, the research results of neural network are remarkable. However, due to the inherent defect that electrical signal is easily interfered and the processing speed is proportional to the energy loss, researchers have turned their attention to light, trying to build neural networks in the field of optics, making full use of the parallel processing ability of light to solve the problems of electronic neural networks. After continuous research and development, optical neural network has become the forefront of the world. Here, we mainly introduce the development of this field, summarize and compare some classical researches and algorithm theories, and look forward to the future of optical neural network.
1 中国民用航空飞行学院 民航安全工程学院，四川 广汉 618307
2 南京理工大学 电子工程与光电技术学院，江苏 南京 210094
在深空探测中探测信号经过对流层延迟后在接收机端信号将出现一定程度的时延，影响探测精度。现有方法主要通过网格模型、空间模型实现时延预测，但由于区域差异导致模型准确度受限，预测精度仍有改进空间。提出了一种基于自适应多输入多输出（MIMO）信号的深空探测对流层延迟预测模型。基于单一收发天线模拟卫星信号MIMO传输方式，然后构建自适应卡尔曼滤波器，通过自适应调整MIMO信号分量权重系数的方法选取最优传输路径以实现对流层延迟量的预测。参与测量的卫星数目为4颗，在不同信噪比以及改变MIMO通道数目情况下开展实验，研究自适应MIMO模型的准确度和实际测量误差。实验结果表明，新方法相对于GPT2模型、GPT2w模型以及实时导航定位中常用的UNB3模型、EGNOS模型的预测精度有较大提高。
深空探测 对流层延迟 自适应卡尔曼滤波 deep space exploration tropospheric delay MIMO MIMO adaptive Kalman filter 红外与激光工程
2020, 49(5): 20190471
Author Affiliations
Abstract
Jiangsu Key Laboratory of Spectral Imaging and Intelligence Sense, Nanjing University of Science and Technology, Nanjing 210094, China
This Letter presents a simple and effective method to improve the signal-to-noise ratio (SNR) of compressing imaging. The main principles of the proposed method are the correlation of the image signals and the randomness of the noise. Multiple low SNR images are reconstructed firstly by the compressed sensing reconstruction algorithm, and then two-dimensional time delay integration technology is adopted to improve the SNR. Results show that the proposed method can improve the SNR performance efficiently and it is easy to apply the algorithm to the real project.
110.1758 Computational imaging 100.3010 Image reconstruction techniques Chinese Optics Letters
2017, 15(10): 101101
南京理工大学电子工程与光电技术学院，江苏 南京 210094
针对传统的红外图像非均匀性校正方法精度低，易破坏图像细节和边缘等缺点，本文提出了一种新的基于全变分理论的红外图像非均匀性校正方法。在分析不同正则项对全变分模型去噪性能影响的基础上，针对红外图像条纹非均匀性的几何特征，对原有的全变分模型进行了修正，使新模型既能约束图像水平方向的梯度，又能保护图像垂直方向的梯度。通过Split Bregman 迭代最小化新的全变分模型，显著降低了计算复杂度，使其能广泛应用于实时视频序列。通过不同环境下对真实场景的实验，表明该方法不但能有效地校正红外图像的条纹非均匀性，还能较大程度地保护住图像的细节和边缘信息。
红外焦平面阵列 非均匀性校正 全变分 Split Bregman 迭代 L1 正则 infrared focal plane array nonuniformity correction total variation Split Bregman iteration L1-norm
Author Affiliations
Abstract
A novel numerical algorithm is proposed to reconstruct the Laplacian of an object field from one single in-line hologram. This method uses two different reconstruction distances of z and z+\Delta z, or two different reconstruction wavelengths of \lambda and \lambda + \Delta to reconstruct one digital in-line hologram. Theoretical analysis shows that when the value of \Delta z or \Delta \lambda is sufficiently small, the difference of the two reconstructed fields is an approximation to the second-order Laplacian differentiation of the object wave, and the zero-order and "twin-image" noise can be almost eliminated simultaneously. Computer numerical simulations and optical experiments are carried out to validate the effectiveness of this algorithm.
090.1995 Digital holography 050.1970 Diffractive optics 110.3010 Image reconstruction techniques 100.2980 Image enhancement Chinese Optics Letters
2012, 10(s1): S10901
Author Affiliations
Abstract
National Defense Key Laboratory of Optoelectronic Engineering, Nanjing University of Science and Technology, Nanjing 210094, China
To decrease the performance difference between the actual microscanning thermal imager and the theoretical value, a germanium lens (placed at a certain angle between the infrared focal plane array and infrared lens) dip angle model of flat optical component microscanning is introduced in this letter. The model is the basis for choosing the dip angle of the germanium lens, which is used in the microscanning thermal imager. In addition, the actual dip angle of the germanium lens is chosen according to the model, the infrared lens parameters of the thermal imager, and the germanium lens parameters of manufacture and installation. Only in this manner can the optimal performance of the microscanning thermal imager based on the flat optical component be obtained. Results of the experiments confirm the accuracy of the conclusions above.
光学 微扫描 平板光学元件 锗片倾斜角度模型 230.3990 Micro-optical devices 100.6640 Superresolution 110.4190 Multiple imaging 230.4040 Mirrors Chinese Optics Letters
2011, 9(5): 052302
