水下成像的现状和发展动向

本文以下水成像在军事，海洋开发工程应用为基础，论述了水下观测和成像中采用微光电视摄像机和激光器件及其系统的技术现状和发展动向。     

Research progress in terahertz quantum cascade lasers

The terahertz(THz)quantum cascade laser(QCL)is a coherent THz source based on intersubband transitions of unipolar carriers.As semiconductor lasers,THz QCLs are solid-state,electrically pumped,tunable,and can be easily integrated.This paper reviews the state of the art for the design,operation temperature,and output power of THz QCLs.Their applications in THz communication and imaging are also discussed.     

Diffusion tensor interpolation profile control using non-uniform motion on a Riemannian geodesic

Tensor interpolation is a key step in the processing algorithms of diffusion tensor imaging (DTI), such as registration and tractography. The diffusion tensor (DT) in biological tissues is assumed to be positive definite. However, the tensor interpolations in most clinical applications have used a Euclidian scheme that does not take this assumption into account. Several Rie-mannian schemes were developed to overcome this limitation. Although each of the Riemannian schemes uses different metrics, they all result in a ‘fixed’ interpolation profile that cannot adapt to a variety of diffusion patterns in biological tissues. In this paper, we propose a DT interpolation scheme to control the interpolation profile, and explore its feasibility in clinical applications. The profile controllability comes from the non-uniform motion of interpolation on the Riemannian geodesic. The interpolation experiment with medical DTI data shows that the profile control improves the interpolation quality by assessing the reconstruction errors with the determinant error, Euclidean norm, and Riemannian norm.     

虚拟红外成像器件的模拟输出电路设计

虚拟红外成像器件是能输出红外图像数据源的器件,在调试图像采集系统的硬件和软件时能替代非制冷红外焦平面阵列。虚拟成像器件的输出具有很好的线性度和精度,可作为图像采集电路的标定器件。文设计了虚拟红外成像器件的模拟输出电路,根据UL03401性能参数,设计了信号调理电路,通过在FPGA上使用ROM中查表方式校正模拟端口输出误差,在实际应用中表现了良好的精确性。     

对偶四元数线阵遥感影像几何定位

提出以对偶四元数为数学工具进行线阵CCD(电荷耦合元件)遥感影像几何定位的全新技术方法。利用对偶四元数建立遥感通用传感器严密成像模型,将光线束的位置和姿态统一用对偶四元数表示,通过传感器扫描光线在空间中的螺旋运动,实现像点到其对应地面点物方坐标的变换,从而克服了成像几何参数(外方位元素)之间的强相关性。按照空间刚体变换线性蒙皮混合理论,可以把刚体变换矩阵分解为平移和旋转两个部分,对平移部分进行线性插值,对旋转部分进行球面插值,从而实现线阵CCD遥感影像外方位元素的解算。按照所建立的成像几何模型,利用某地区Geoeye-1遥感影像进行几何定位实验,实验结果表明新算法获得的几何定位精度优于传统算法,能够解决定位参数之间的相关性问题。     

Bayesian network based general correspondence retrieval method for depth sensing with single-shot structured light

This study reinvestigated one of the most fundamental problems in structure light depth sensing field: correspondence retrieval of features between patterns and images. We formulate the global optimum correspondence retrieval by maximizing a conditional probability of correspondence given observed features, which is depicted by a Bayesian network. Different from traditional “code-only” based correspondence retrieval methods, the proposed Bayesian network based method exploits the positional correlations of correspondences of neighboring features, namely, the correspondences of poorly detected features are estimated with the aid of the correspondences of well detected features. The method performs especially well on challenging scenes with rich depth variations, abrupt depth changes, edges, etc. Experiments show that the proposed method increase the correspondence accuracy by about 40% on challenging scenes, compared with traditional “code-only” based correspondence retrieval methods.     

Film and television industry cloud exhibition design based on 3D imaging and virtual reality

The three-dimensional virtual scene can provide users with a visual three-dimensional virtual environment, with various multimedia channels such as sound, video, force feedback equipment, etc., to bring users a completely immersive interactive experience. This paper introduces 3D imaging and virtual reality technology in the film and television industry cloud exhibition, and develops a virtual display platform. First of all, this paper divides the registration into two processes: camera calibration and joint calibration of the camera and laser scanner based on the calibration results. Camera calibration can determine the plane model of the calibration board in the camera coordinate system, and the joint calibration uses the RANSAC algorithm to extract the point cloud of the plane model of the calibration board, and then optimizes the distance between the points in the plane model point cloud and the corresponding plane in the camera coordinate system Find the optimal transformation between the two sets of data, and then calculate the registration relationship between the point cloud and the image. Secondly, this article conducts a demand analysis of the film and television industry cloud exhibition platform based on virtual reality technology, including the business goals set by the platform, platform system analysis, overall design, and system operating environment and configuration requirements. This model provides a feasible solution for the visual interaction of the cloud exhibition design of the film and television industry.     

ASRNet: Adversarial Segmentation and Registration Networks for Multispectral Fundus Images

Multispectral imaging (MSI) technique is often used to capture images of the fundus by illuminating it with different wavelengths of light. However, these images are taken at different points in time such that eyeball movements can cause misalignment between consecutive images. The multispectral image sequence reveals important information in the form of retinal and choroidal blood vessel maps, which can help ophthalmologists to analyze the morphology of these blood vessels in detail. This in turn can lead to a high diagnostic accuracy of several diseases. In this paper, we propose a novel semi-supervised end-to-end deep learning framework called “Adversarial Segmentation and Registration Nets” (ASRNet) for the simultaneous estimation of the blood vessel segmentation and the registration of multispectral images via an adversarial learning process. ASRNet consists of two subnetworks: (i) A segmentation module S that fulfills the blood vessel segmentation task, and (ii) A registration module R that estimates the spatial correspondence of an image pair. Based on the segmention-driven registration network, we train the segmentation network using a semi-supervised adversarial learning strategy. Our experimental results show that the proposed ASRNet can achieve state-of-the-art accuracy in segmentation and registration tasks performed with real MSI datasets.