1 国防科技大学电子科学学院, 湖南长沙 410073
2 怀化学院电气与信息工程学院, 湖南怀化 418008
深度学习(DL)在语音识别、图像物体识别上取得了卓越的成效, 深度学习代替传统处理技术, 成为了研究该领域的主要处理方法。在雷达领域, 深度学习用于雷达目标识别和分类, 也取得了很好的效果, 进而, 人们试图将深度学习用于雷达成像。本文根据近几年所公开的文献资料, 按照雷达成像的特点, 分类介绍深度学习用于雷达成像的研究进展; 之后, 对深度学习用于雷达成像的可行性、样本选取、泛化以及成像质量的评价等开放性问题提出了作者的设想, 并对深度学习用于雷达成像进行了展望。
深度学习 雷达成像 可行性 样本 泛化 Deep Learning radar imaging feasibility samples generalization 太赫兹科学与电子信息学报
2023, 21(9): 1086
红外与激光工程
2021, 50(12): 20210790
集成成像较小的再现深度一直都限制着集成成像的发展和应用,针对此问题提出了一种增大集成成像再现深度的方法。该方法在微透镜阵列与显示屏之间附加一个光孔阵列,利用光孔阵列限制显示屏上像素发出光线的发散角,从而有效地增大集成成像的再现深度。对集成成像的再现原理进行了深入分析,讨论了光孔直径与集成成像再现深度的关系。采用ASAP光学模拟软件对所提方法和传统方法进行了模拟对比实验,实验结果显示当光孔直径占透镜元节距的64%时,所提方法的再现深度是传统集成成像再现深度的1.5倍,实验结果验证了理论推导的正确性。
成像系统 再现深度 光孔阵列 微透镜阵列 三维显示
Author Affiliations
Abstract
A crosstalk-free integral imaging display consisting of a display panel and double plano-convex micro-lens array is proposed. The double plano-convex micro-lens array includes two micro-lens arrays, A and B. Micro-lens array A is used to eliminate crosstalk by completely reflecting crosstalk lights. Micro-lens array B, located near micro-lens array A, is used to display three-dimensional images. Computer simulations based on ray-tracing are conducted. Crosstalk-free reconstruction images may be clearly observed from the simulation results.
110.6880 Three-dimensional image acquisition 100.6890 Three-dimensional image processing 110.2990 Image formation theory Chinese Optics Letters
2013, 11(6): 061101
Author Affiliations
Abstract
We propose a one-dimensional integral imaging (1DII) display that consists of a display panel and a gradient-aperture parallax barrier. The gradient-aperture parallax barrier is symmetrical, and its slit widths gradually increase from both sides to the middle. The leftmost and rightmost slits are used to fix the viewing angle, whereas the other slits are used to increase the optical efficiency. A prototype of the proposed 1DII display is developed. Its optical efficiency is higher than that of the conventional display, but the viewing angles are the same.
110.0110 Imaging systems 110.6880 Three-dimensional image acquisition 230.0230 Optical devices Chinese Optics Letters
2013, 11(7): 071101
Author Affiliations
Abstract
One-dimensional (1D) integral imaging based on parallax images' virtual reconstruction is proposed. The 1D integral imaging contains parallax images' capture process, parallax images' virtual reconstruction process, and 1D elemental image array's generation process. A pixel mapping algorithm is deduced to implement the last two processes; a 1D elemental image array is generated by the mapping of pixels on the parallax images obtained using a 1D camera array. The proposed 1D integral imaging can capture the 1D elemental image array of a real three-dimensional (3D) scene.
110.6880 Three-dimensional image acquisition 100.6890 Three-dimensional image processing 110.2990 Image formation theory Chinese Optics Letters
2013, 11(4): 041101