十亿像素瞬态成像系统实时图像拼接

为了满足工程应用对图像拼接实时性的要求,依据已设计完成的基于同心球透镜与微相机拼接阵列复合结构的十亿像素瞬态成像系统,提出一种基于统一计算设备架构(CUDA)与先验信息相结合的自适应图像拼接并行加速算法。首先,利用高精度四维标定平台对相邻微相机成像重叠区域进行预标定。接着,采用基于CUDA的快速鲁棒特征(SURF)方法检测提取重叠区域图像的候选特征点集。然后,运用基本线性代数运算子程序(CUBLAS)加速基于随机KD-Tree索引的近似最近邻搜索(ANN)算法,用于获取初始匹配点对。最后,提出一种改进的并行渐近式抽样一致性(IPROSAC)算法,用于剔除误匹配点对和空间变换矩阵的参数估计,从而得到拼接图像的空间几何变换关系。实验结果表明,该算法的图像拼接时间为287 ms,与单独采用CPU串行算法相比速度提高了近30倍。     

Image Encryption Scheme with Compressed Sensing Based on a New Six-Dimensional Non-Degenerate Discrete Hyperchaotic System and Plaintext-Related Scrambling

Digital images can be large in size and contain sensitive information that needs protection. Compression using compressed sensing performs well, but the measurement matrix directly affects the signal compression and reconstruction performance. The good cryptographic characteristics of chaotic systems mean that using one to construct the measurement matrix has obvious advantages. However, existing low-dimensional chaotic systems have low complexity and generate sequences with poor randomness. Hence, a new six-dimensional non-degenerate discrete hyperchaotic system with six positive Lyapunov exponents is proposed in this paper. Using this chaotic system to design the measurement matrix can improve the performance of image compression and reconstruction. Because image encryption using compressed sensing cannot resist known- and chosen-plaintext attacks, the chaotic system proposed in this paper is introduced into the compressed sensing encryption framework. A scrambling algorithm and two-way diffusion algorithm for the plaintext are used to encrypt the measured value matrix. The security of the encryption system is further improved by generating the SHA-256 value of the original image to calculate the initial conditions of the chaotic map. A simulation and performance analysis shows that the proposed image compression-encryption scheme has high compression and reconstruction performance and the ability to resist known- and chosen-plaintext attacks.     

基于特征点及优化理论的图像自动拼接方法

提出了一种新的图像拼接方法,首先利用相位一致性(phase congruency)算法进行特征点检测,利用本文提出的匹配点优选策略进行特征点对自动选取,然后用LM(Levenberg-Marquardt)算法进一步优化变换矩阵,最后对拼接结果进行融合处理,获得无缝拼接的图像.该方法把基于特征点和基于优化理论的拼接方法有效相结合,且能充分利用图像重叠部分的信息,在一定程度上克服了噪声及光照不均的影响,较传统方法具有更强的鲁棒性和更高的拼接精确度.试验结果证明了该方法的有效性.     

The formation of large scale reconstructed images through the use of holograms

We propose and demonstrate a method which constructs large size multicolor images using a holographic technique in limited spaces, i.e., tunnels. A traffic sign image of the same size and color as an actual traffic sign board is displayed using many comparative small holograms. Each hologram is composed of a single computer generated hologram (CGH). Though it was necessary to make many CGHs, we have shortened the hologram production time by putting a reconstructed image on long distance, and using many reconstruction sources. Moreover, though cross-talk images are caused in a color reconstruction, by adjusting the optical system it is possible to make sure that drivers see only the appropriate image. Many holograms function as one big hologram, and it has been confirmed that a large reconstructed image can be displayed in a limited space.     

复眼式光学成像系统畸变测量与校正技术研究

复眼式光学成像系统在大视场侦查、图像识别、目标探测等领域较传统单孔径光学系统优势突出,但随着视场的增加,子孔径本身的成像畸变及多个子孔径的安装位置误差引起的畸变会直接影响拼接图像的质量。针对该问题,采用光电测量技术对复眼系统进行畸变测量与校正,生成多模动态电子畸变测量靶标,构建畸变测量校正模型,建立多项式拟合算法,采用最小二乘法获得畸变系数,通过双线性插值法模型对图像进行重建。实验结果表明,校正后的平均相对畸变优于0.1%,满足大视场复眼式光学成像系统的畸变校正和图像拼接的精度要求。     

机械快门滤光片式多光谱成像仪相对定标方法研究

大面阵滤光片阵列多光谱成像仪采用了全帧转移大面阵探测器,并利用机械快门保证采集数据时不发生拖尾现象,但由于机械快门开合时间的不稳定性,使多光谱成像仪数据出现区域间的亮度不均匀,而且还会造成彩色图像偏色条带。故提出了一种基于目标辐射一致性的相对辐射定标方法,利用图像重叠部分的辐射特性,计算光谱图像的辐射矫正因子,对拼接图像的非均匀性进行矫正。并利用差分评价矫正效果。最后对多光谱相机采集的外场实验数据进行处理和矫正,结果表明,该方法可以很好的矫正单色拼接图像区域间的非均匀性,并可以消除彩色合成图像中的偏色条带;适用于有重叠部分的图像拼接,并且对于拼接误差不敏感。     

多视角图像标定与图像合成

近年来随着视觉成像技术的快速发展，成像系统也在不断改进。单一摄像机成像已经不能满足人们对大视场高动态的需求，所以具有大视场优势的多摄像机系统引起广泛的关注。依据复眼成像原理，使用多摄像机系统研究其成像特性。考虑到成像过程中的损耗和镜头的特性，对所用摄像机进行辐照度定标，并进行光晕校正，同时求解出了摄像机辐射响应和暗电流的函数，随后对多摄像头系统进行大视场图像拼接和高动态范围成像。实验表明，辐射定标对于上述成像具有良好效果。     

大尺寸物光波面彩色数字全息高质量重建研究

为解决检测面尺寸较大时CCD难以得到高质量数字全息图的问题,本文利用负透镜设计光学系统让CCD接收来自物体的缩小虚像,以球面波为参考波,使用单色CCD近距离得到三种色光照射下的大尺寸彩色物体的数字全息图,然后采用可控放大率波面重建算法得到同一尺寸的数字全息重建像,合成彩色数字全息重建像.同时,使用两种消零级方法去除零级干扰,提高重建像质量,一种方法利用空间光调制器相移技术在参考光中加入一次任意相移,记录两幅数字全息图,消除重建零级像|另一种方法使用“无干扰全息图”消除重建零级像及共轭像.本文讨论结果可为大物体彩色数字全息及多波长数字全息检测应用提供有益的参考.     

基于变换矩阵的BGA X-ray图像倾斜识别

针对基于X-Ray的焊球阵列封装(BGA)焊接缺陷检测中,现有的图像处理算法无法有效处理PCB意外倾斜的问题,提出了一种自动判断BGA焊点X射线图像倾斜程度并进行校正的方法。首先根据BGA焊点阵列排布的特性恢复出原始图像中的结构信息,然后使用广义逆矩阵求解出理想正视图像与实际倾斜图像之间的变换矩阵,最后对倾斜图像进行逆变换以得到理想正视图像,并使用变换矩阵估计出PCB的倾斜程度。实验针对正视与倾斜的BGA焊点的X射线图像,运用所提方法,准确判别出图像的倾斜程度,并进行了校正。此方法既可以作为一种图像质量评价算法,在BGA X-Ray缺陷检测中判断图像是否为需要的正视图像;也可以作为一种图像自动校正算法,提升基于X射线的自动检测系统的适应性。     

一种基于相位相关的亚像素红外图像配准算法

图像匹配技术广泛应用于各种图像处理任务,如图像拼接、机器视觉等。通常匹配算法的精确度只能达到像素级别,但在很多图像处理任务如超分辨率重建中需要亚像素精度的图像配准。提出了一种基于相位相关的亚像素图像配准算法。根据两幅离散数字图像的相位相关矩阵中的最大值以及其附近若干点可以拟合估计出实际的峰值位置,进而实现两幅图像的亚像素运动估计。提出的算法针对热像仪采集的红外图像进行匹配实验,实验结果表明该算法精度相比通常的亚像素匹配算法较高,且具有更好的实用性。     

Vessel boundary extraction based on a global and local deformable physical model with variable stiffness

Reliable and efficient vessel cross-sectional boundary extraction is very important for many medical magnetic resonance (MR) image studies. General purpose edge detection algorithms often fail for medical MR images processing due to fuzzy boundaries, inconsistent image contrast, missing edge features, and the complicated background of MR images. In this regard, we present a vessel cross-sectional boundary extraction algorithm based on a global and local deformable model with variable stiffness. With the global model, the algorithm can handle relatively large vessel position shifts and size changes. The local deformation with variable stiffness parameters enable the model to stay right on edge points at the location where edge features are strong and at the same time, fit a smooth contour at the location where edge features are missing. Directional gradient information is used to help the model to pick correct edge segments. The algorithm was used to process MR cine phase-contrast images of the aorta from 20 volunteers (over 500 images) with excellent results.     

Infrared Image Series Automatic Registration and a History & Tendency Analysis

Infrared camera has been widely used in electrical power systems for predictive and preventive thermal fault diagnosis. Without historical IR images being considered, most thermograph analysis used at present are concerned with static image. The dynamic history and tendency analysis which compares the newly taken image with historical images is particularly attractive. Conclusions can be drawn from temperature variations rather than just from anomalous thermal pattern. However the spatial distortions prevent such dynamic analysis from being practicable. So the need to register infrared images arises. This paper proposes an automatic registration algorithm based on closed temperature contour matching for infrared image series of a target equipment taken at different times. Centroid Distance Function is used in image registration for the first time. The algorithm can remove spatial distortions, bring temperature variations into prominence and endow infrared camera with dynamic analysis ability. Experiment shows that the algorithm is simple, effective and extensible.     

Analysing the sintering of heterogeneous powder structures by in situ microtomography

This work analyses the microstructure changes of various copper-based powder systems during sintering from 3D images provided by in situ synchrotron microtomography. The investigated systems include a copper powder with a wide particle size distribution of 0–63 µm poured into a quartz capillary, a pre-sintered copper compact with artificially created large pores and a mixture of copper and alumina particles. The experiments were carried out at the European Synchrotron in Grenoble, France. Powders were sintered up to 1060°C under reducing atmosphere in a furnace located between the X-ray source and the detector. During each experiment, 3D images were taken at various times of the thermal cycle. We have obtained images with a resolution of 1.5 µm and the time of acquisition of every image was ～1 min. Quantitative analysis of these images allowed the changes of various important parameters to be followed. Such parameters characterise the sintering process at the particle length scale: interparticle coordination, pore size distribution and particle centre-to-centre distance. Moreover, by tracking the displacement of each particle centre and comparing it to the displacement predicted by classical mean field assumption, we have been able to assess the magnitude of particle rearrangement occurring during sintering. From these data, the sintering behaviour of heterogeneous powder systems is discussed with particular emphasis of collective particle phenomena.     

Novel ultrasonic real-time scanner featuring servo controlled transducers displaying a sector image

This paper describes a new real-time servo controlled sector scanner that produces high resolution images and has functionally programmable features similar to phased array systems, but possesses the simplicity of design and low cost best achievable in a mechanical sector scanner. The unique feature is the transducer head which contains a single moving part--the transducer--enclosed within a light-weight, hand held, and vibration free case. The frame rate, sector width, stop action angle, are all operator programmable. The frame rate can be varied from 12 to 30 frames s-1 and the sector width from 0 degrees to 60 degrees. Conversion from sector to time motion (T/M) modes are instant and two options are available, a freeze position high density T/M and a low density T/M obtainable simultaneously during sector visualization. Unusual electronic features are: automatic gain control, electronic recording of images on video tape in rf format, and ability to post-process images during video playback to extract T/M display and to change time gain control (tgc) and image size.     

基于单目结构光的大物体三维测量关键方法的研究

在单目结构光的三维测量系统中,由于投影仪倾斜投影, 参考平面上的条纹周期展宽, 给测量带来误差,降低了测量精度。同时受大物体自身几何和形貌等因素的影响,以及相交轴测量系统的限制,其单幅面测量范围受限,很难一次测量大物体完整的三维形貌,而且在测量大物体时,摄像机镜头非线性畸变也影响测量精度。根据参考平面上光栅条纹的周期变化规律,提出了一种适用性好、方便快捷的条纹周期校正的理论模型, 在此基础上, 提出了基于条纹周期校正的四步相移法的理论模型,进而提出了基于条纹周期校正的时间相位展开法的理论模型。采用摄像机镜头非线性畸变校正模型,提高测量精度。在被测物表面粘贴标志点,获取其三维坐标,利用SVD分解和L-M优化算法求取转换矩阵,并在设定的全局坐标系下实现三维图像拼接,采用线性加权算法,对重叠区域进行图像融合。实验结果表明,X轴的拼接误差为0.14 mm,Y轴的拼接误差为0.16 mm,Z轴的拼接误差为0.19 mm,其拼接误差均在测量误差允许范围之内。     

基于神经网络的三维宽场显微图像复原研究

提出一种利用BP神经网络进行三维宽场显微图像复原的非线性映射方法,将三维图像转化为二维图像进行处理,利用神经网络的学习能力,通过训练,建立含有散焦信息的二维模糊图像与二维清晰图像之间的映射关系,然后对切片堆叠进行逐幅复原,从而实现显微图像的三维复原.得到的复原图像在视觉上和定量分析上都获得了很好的效果.由于采用小规模神经网络,训练时间短,计算量小,使实时复原成为可能.     

基于像素结构空间光调制器的全息再现像问题研究

基于纯相位空间光调制器的全息显示系统在重构显示时,再现像的视觉效果受到空间光调制器像素结构引起的多级衍射光和多级再现像的干扰。在分析具有有限填充因子空间光调制器的像素结构对再现像影响的基础上,提出了一种提高全息再现像的视觉效果并且再现像成像位置和大小可调节的方法。先加载闪耀光栅到纯相位全息图,其次通过叠加会聚球面波相位,分离再现像与空间光调制器像素结构引起的多级衍射光的聚焦平面的位置,再利用光阑和高通滤波器的共同作用,消除高级衍射光、多级再现像以及零级光干扰对重构视觉效果的影响,最后引入成像透镜,调节再现像的成像位置与大小。建立了一套基于硅基液晶的全息显示系统用于实验验证。实验结果表明,最终的单一再现像清晰且可以方便地调节成像位置和大小。该方法同样适用于各种基于像素结构空间光调制器的全息光学系统。     

神龙一号参数测量用高速两分幅相机

采用长焦距镜头的后工作空间全口径分光原理,利用门控型像增强器、CCD相机、基于大规模可编程集成电路的高速快门控制触发系统等部件,研制了具有较高时间分辨能力和高灵敏度的两分幅高速相机,并在此基础上建立了束参数的高速测量系统。两分幅相机的最高快门速度约3 ns,幅间间隔时间则具有以0.5 ns的步进进行调节的能力;快门时间及幅间间隔时间可以分别独立调节,最大可到1 s;同时具有较好的线性度和空间响应的均匀性,等效背景噪声低到约5 electronspixel-1s-1,并且分幅相机灵敏度调节范围大。该系统一次可以拍摄两幅图像,图像阵列可达到1 0241 024,满足神龙一号的各种测量要求。     

三维显微图像复原及点扩散函数的研究

三维显微图像去卷积计算量巨大,运算时间长。评价和分析了三维点扩散函数、散焦像、光学切片以及不同三维点扩散函数去卷积的复原图像,进一步证明了散焦光信息对焦面像的干扰,主要来自于焦面像两侧附近的散焦像。提出了折中选择三维PSF的空间大小,可以获得良好的复原效果,并且减少运行时间。提出了频谱图均值的新概念,作为评价图像清晰及模糊程度的标准,并运用于散焦像、光学切片以及复原结果图的评价。