Virtual viewpoints reconstruction via Fourier slice transformation

We propose a new method of calculating the arbitrary viewpoints for auto‐stereoscopic display. The three‐dimensional (3D) object is first virtually reconstructed in 3D spaces by mapping each pixel with a depth according to the depth mapping. We then calculate the Fourier spectrum of the 3D object by the fast Fourier transformation. The arbitrary viewpoints are reconstructed by “slicing” the 3D Fourier spectrum. To repair “black hole” artifacts, the regions in the background are calculated by advanced boundary in‐painting. Experimental results show the effectiveness and accuracy of the proposed algorithm to calculate viewpoints with arbitrary viewing angles. A comparison is also presented, which indicates that the proposed algorithm is more accurate than conventional method, and the advanced boundary in‐painting can save three quarters of time than the conventional in‐painting method.     

Enhanced depth of field of computer-generated holographic stereograms by twice close-range capturing with occlusion processing

In pinhole integral imaging (PII)-based hologram, each pinhole is regarded as a point light source to emit light rays in different directions to simulate light propagation from the reconstructed 3D information to the computer-generated hologram (CGH). The PII-based hologram presents at a working mode similar to resolution-priority integral imaging (RPII). The depth of field (DOF) is inevitably limited. In this paper, we use twice close-range capturing to enhance the DOF of this hologram. Twice close-range capturing can ensure that the virtual objects with different depths are not far from their respective image reference planes, and the detailed information of the back object also can be preserved. Besides, we use a simple ray-tracing method to solve the occlusion problems. Numerical and optical reconstruction verified that the proposed method can reconstruct clearer 3D images compared with the single capturing method. In addition, the proposed method can present a natural occlusion effect. Thus, the DOF is enhanced.     

保持特征的点云自适应网格重建

由于3维扫描点云通常存在噪音和缺失数据,提出了一种鲁棒的点云网格重建算法。对张量矩阵方法估计的点云法向进行增强特征处理,在频域中进行3维快速傅里叶变换,提取粗糙离散等值面。原始点云经梯度方向迭代移动后,过滤噪音和剔除离群点,并修补点云缺失数据。点云被自适应筛选后,利用圆球相交的方法生成新的三角形。实验表明,该算法具有快速、稳定可靠和内存消耗小的优点。     

Special purpose computer system with highly parallel pipelines for flow visualization using holography technology

We have designed a PC cluster system with special purpose computer boards for visualization of fluid flow using digital holographic particle tracking velocimetry (DHPTV). In this board, there is a Field Programmable Gate Array (FPGA) chip in which is installed a pipeline for calculating the intensity of an object from a hologram by fast Fourier transform (FFT). This cluster system can create 1024 reconstructed images from a 1024×1024-grid hologram in 0.77 s. It is expected that this system will contribute to the analysis of fluid flow using DHPTV.     

基于相关型图像传感器3维人脸成像的3维AAMs人脸识别方法的研究

主动外观模型是基于统计分析建立物体2维模型的有效方法,它融合了目标的形状和纹理信息。在基于相关型图像传感器3维人脸成像的基础上,提出了一种建立3维人脸模型的方法,该方法利用由相关型图像传感器得到的深度信息和与之对应的亮度信息将2维AAMs扩展为3维AAMs,融合人脸的形状,纹理和深度信息来构建3维人脸模型。人脸识别实验结果表明,该方法在不同人脸姿态,表情和光照条件下识别效果要优于Eigenface和2维AAMs。     

Modification of a two-dimensional fast Fourier transform algorithm with an analog of the Cooley–Tukey algorithm for image processing

Two-dimensional fast Fourier transform (FFT) for image processing and filtering is widely used in modern digital image processing systems. This paper concerns the possibility of using a modification of two-dimensional FFT with an analog of the Cooley–Tukey algorithm, which requires a smaller number of complex addition and multiplication operations than the standard method of calculation by rows and columns.     

一种偶数基Cooley-Tukey FFT高性能实现方法

快速傅里叶变换(Fast Fourier Transform,FFT)是最重要的基础算法之一,在科学计算、信号处理、图像处理等领域都有着广泛的应用。随着这些应用领域对实时性需求的进一步提高,FFT算法面临着越来越高的性能要求。在现有的FFT算法库中,FFT算法的求解速度和计算精度受到一定程度的限制,而且也少有研究者对偶数基Cooley-Tukey FFT的高性能实现提出相应的优化策略并对技术进行深入研究。基于此,文中提出了一套针对偶数基的Cooley-Tukey FFT的优化策略和方法。首先构建一个SIMD(Single Instruction Multiple Data)友好、支持混合基的蝶形网络,然后根据偶数基旋转因子特性最大限度地降低蝶形计算的复杂度,接着通过SIMD汇编优化、汇编指令重排及选择、寄存器分配策略制定、高性能矩阵转置算法等方法来优化应用,最后实现一个高性能的FFT算法库。目前,最流行、应用最广的FFT有FFTW和Intel MKL。实验结果表明,在X86计算平台上,新提出的这套针对偶数基Cooley-Tukey FFT的技术所实现的FFT算法库的性能全面优于MKL和FFTW。所提出的这套高性能算法优化和实现技术体系,可推广到除偶数基以外的其他基的实现和优化上,为进一步的研究开发工作奠定一定的基础,进而突破FFT算法在硬件平台上的性能瓶颈,实现一套针对特定平台的高性能FFT算法库。     

基于DMFT的高动态载波捕获算法的改进

基于FFT的载波捕获方法对高动态信号不能适用,离散匹配傅里叶变换(DMFT)虽可用于高动态信号,但是其运算量大、精确度差。基于以上分析本文提出了将延迟自相关、FFT与DMFT三者相结合的二维载波捕获算法。首先将中频采样信号与其延迟做自相关,通过信号的延迟自相关的FFT得到频率变化率的粗略估计值,进而得到起始频偏的粗略估计值,然后在所得值附近利用DMFT进行搜索,从而获得高精度的参数估计值。此方法缩小了搜索的范围,在运算量减少的同时,也提高了参数的估计精度。仿真结果证明本文提出的方法有效可行。     

基于谱域特征提取与线性回归分类的智能人脸识别算法

针对人脸识别中由于光线、表情变化和遮挡导致人脸图像变化的问题,提出了一种谱域特征提取与线性回归分类算法相结合的智能人脸识别方法。为了实现特征提取的目的,首先使用Viola-Jones算法从原始图像中提取初始人脸部分,并将其转换为120×120像素大小的灰度图像;然后提出了一种计算极坐标傅里叶变换（FFT）以获得预处理人脸图像主要幅度谱特征的新框架,进一步在预处理的图像上执行2D-DFT,并表示为1D P-FFT。特征值是1D P-FFT幅值中的最大值,提取的特征值用于构造表示人脸图像的符号对象。最后利用快速有效的线性回归分类算法实现分类。在AR和GT数据库上进行了各种实验,分别取得了97.51％和98.02％的准确率,与最近报道的一些人脸识别技术相比,提出的方法识别准确率更高。     

基于FPGA的二维FFT图像边缘增强设计

在处理图像类信息时,图像细节往往能传达更多信息,是人们较为关注部分。针 对在光照不理想的条件下,传感器采集到的图像对比度低、细节难以分辨的问题,提出一种基 于现场可编程门阵列(FPGA)的二维快速傅立叶变换的图像边缘提取及增强方法。通过模块化设 计,完成 4 路并行 512×512 点快速傳里叶变换(FFT)运算处理器设计,并通过 FFT 模块复用减 少 FPGA 内资源消耗,同时实现图像频谱的高通滤波算法及傅立叶逆变换算法。经过仿真与实 验,确定该方法有效可靠,实时性强,可以满足工业上图像处理的需求。     

分数傅里叶全息图的快速算法及数字重现

论文通过分析菲涅耳衍射积分的快速算法,提出了一种基于快速傅里叶变换的分数傅里叶变换的数值模拟算法,并研究了基于此快速算法的分数傅里叶变换全息图的计算机生成及数字重现。     

An approach for refocusing of ground fast-moving target and high-order motion parameter estimation using Radon-high-order time-chirp rate transform

Long synthetic aperture time can improve the imaging quality of a ground moving target, whereas a moving target may be severely smeared in the cross-range image due to the range migration and the Doppler frequency migration. In this paper, the effects of the third-order Doppler broadening and Doppler ambiguity of a fast-moving target are considered. To address these issues, a novel motion parameter estimation method named high-order time-chirp rate transform (HTRT) is proposed, and then a new synthetic aperture radar (SAR) imaging method based on Radon-HTRT (RHTRT) for a ground moving target is developed. The major contributions are as follows: 1) The proposed SAR imaging method can eliminate the Doppler ambiguity effect. 2) The proposed method can realize longer time coherent integration than Radon–Fourier transform (RFT) and Radon–fractional Fourier transform (RFRFT) methods. 3) The proposed method is computationally efficient since HTRT can obtain the motion parameters of a moving target via performing the 2-dimensional (2-D) fast Fourier transform (FFT). Both the simulated and real data processing results show that the proposed method can finely image a ground moving target in a high signal-to-clutter and noise ratio (SCNR) environment.     

Integral floating‐image display using two lenses with reduced distortion and enhanced depth

Abstract— An integral floating display (IFD) with a long depth range without floating lens distortion is proposed. Two lenses were used to reduce barrel distortion of the floating lens and three‐dimensional (3‐D) image deformation from object‐dependent longitudinal and lateral magnifications in the floating‐display system, combined with an integral imaging display. The distance between the floating lenses is the sum of their focal lengths. In the proposed configuration, lateral and longitudinal magnifications are constant regardless of the distance of the integrated 3‐D images, so the distortions from the distant‐dependent magnifications of the floating lens do not occur with the proposed method. In addition, the proposed floating system expands the depth range of the integral imaging display. As a result, the display can show a correct 3‐D floating image with a large depth range. Experimental results demonstrate that the proposed method successfully displays a 3‐D image without floating lens distortions across a large depth range.     

Fast implementation of ground maneuvering target focusing

Ground maneuvering target detection is a hot topic in the applications of synthetic aperture radar (SAR), whereas its focusing performance is severely deteriorated by range migration and Doppler frequency migration during a long integration time. This paper proposes a novel method to image the target and estimate its parameters via performing two independent 2-dimensional (2-D) searches after a parameter separation operation. In order to improve the search speed, we set the limited search ranges and propose local mapping sparse Fourier transform (LMSFT) to replace fast Fourier transform (FFT). Compared with the traditional algorithms, the proposed method can realize fast coherent integration of multiple maneuvering targets via compensating the high-order range migration and Doppler frequency migration. In addition, the proposed method is stable under noise. Several simulation results have validated the effectiveness of the proposed method.     

B小波和阶跃谱分析在缺损频谱成像中的应用

提出一种用于频域的快速B小波变换方法和阶跃谱分析理论,并把它们用于磁共振缺损频谱数据中提取高频频谱、恢复完整频谱,重建图像.在此方法中,首先用快速B小波变换方法,从低频磁共振数据中提取、恢复缺少的高频分量的特征信息;然后用阶跃谱分析理论,由这些特征信息构成高频磁共振频谱、恢复完整频谱数据;最后用FFT(fast Fourier transform)重建磁共振图像.实验和模拟结果都表明,重建图像质量优于现有方法.     

基于稀疏贝叶斯学习的雷达目标成像技术

利用傅立叶变换进行雷达目标成像,分辨率受瑞利准则的限制;超分辨成像技术能最著改善雷达图像的分辨率,但算法的复杂性急剧增加并且正则化参数不易选取.以稀疏贝叶斯学习为基础,针对雷达成像系统的结构特点,提出了一种基于快速傅立叶变换(FFT)和分块托普里兹(Toeplitz)系统的快速超分辨成像算法.算法无需存储系数矩阵,极大地降低了存储量和运算量.进一步,通过寻找拟合误差曲线和稀疏性度量函数曲线的交点实现了正则化参数的方便选择.仿真结果表明,算法对雷达目标图像具有良好的分辨率增强能力.     

可逆的DCT整型变换与无失真图像压缩

使用提升的方法,利用FFT(fast Fourier transform)的蝶型构造,完成了FFT与DCT(discrete cosine transform)的从整数到整数的变换.变换本身是可逆的,因此非常适合于无失真图像压缩.     

Depth image‐based rendering for multiview generation

Abstract— Techniques for 3‐D display have evolved from stereoscopic 3‐D systems to multiview 3‐D systems, which provide images corresponding to different viewpoints. Currently, new technology is required for application in multiview display systems that use input‐source formats such as 2‐D images to generate virtual‐view images of multiple viewpoints. Due to the changes in viewpoints, occlusion regions of the original image become disoccluded, resulting in problems related to the restoration of output image information that is not contained in the input image. In this paper, a method for generating multiview images through a two‐step process is proposed: (1) depth‐map refinement and (2) disoccluded‐area estimation and restoration. The first step, depth‐map processing, removes depth‐map noise, compensates for mismatches between RGB and depth, and preserves the boundaries and object shapes. The second step, disoccluded‐area estimation and restoration, predicts the disoccluded area by using disparity and restores information about the area by using information about neighboring frames that are most similar to the occlusion area. Finally, multiview rendering generates virtual‐view images by using a directional rendering algorithm with boundary blending.     

分数傅里叶变换合成计算全息与数字重现

传统的计算全息图大多采用菲涅耳衍射积分得到,但菲涅耳衍射积分在描述整个衍射光场中,没有一个统一的采样方法对积分进行数值计算,从而给计算带来不便。为了更好地研究计算全息图问题,文章引入了分数傅里叶变换,通过利用分数傅里叶变换的一种快速数值模拟算法,提出了一种基于分数傅里叶变换的合成空间三维物体全息图的新方法,并用计算机模拟了合成的全息图及其数字重构的结果。实验结果表明:由于分数阶的引入,得到一种处理光场衍射问题的统一算法,因此用分数傅里叶变换来处理光场衍射问题是十分理想的。     

基于Hough变换和FFT多投影精测的织物纬斜快速检测方法研究

织物纬斜角度检测是布料整纬的关键技术环节,如何快速、准确检测纬斜角度对提高整纬质量具有重要意义;针对现有布料图像整纬方法存在速度慢、精度不高的缺点,提出了基于Hough变换原理和快速傅里叶变换(FFT)多投影的织物纬斜图像快速检测方法;首先对采集到的织物图像通过傅里叶变换对图像进行频域滤波再逆变换,滤除图像中不表示纬斜方向的区域信息,其次使用Sobel边缘方向检测算子对图像进行卷积以得到边缘方向图,提取纬纱方向信息,利用形态学滤波得到纬纱骨架图,进一步精简纬纱区域以减少计算量,最后进行Hough变换和FFT多投影分析得到织物图像的纬斜角度;实验测试证明对于不同类型的织物图像,该算法的检测时间低于0.55 s,误差值低于0.2°,能够兼顾检测精度和检测速度,满足工程实际应用要求.