Distortion Invariant Binary Image Recognition Based on Central Projection Correlation

1　Ｉｎｔｒｏｄｕｃｔｉｏｎ　　ＴｈｅｍａｔｃｈｅｄｓｐａｔｉａｌｆｉｌｔｅｒｉｎｇｐｒｏｐｏｓｅｄｂｙＶａｎｄｅｒｌｕｇｔ[1] ｈａｓｒｅｃｅｉｖｅｄｃｏｎｓｉｄｅｒａｂｌｅａｔｔｅｎｔｉｏｎｂｅｃａｕｓｅｉｔｈａｓｉｎｈｅｒｅｎｔａｄｖａｎｔａｇｅｏｆｓｈｉｆｔｉｎｖａｒｉａｎｃｅ ,ｂｕｔｔｈｉｓｔｅｃｈｎｉｑｕｅｓｕｆｆｅｒｓｆｒｏｍｓｅｎｓｉｔｉｖｉｔｙｔｏｒｏｔａｔｉｏｎａｎｄｓｃａｌｅｄｅｆｏｒｍａｔｉｏｎ .Ｔｈｅｉｎｖａｒｉａｎｃｅｉｓｔｈｅｋｅｙｔｏｐａｔｔｅｒｎｒｅｃｏｇｎｉｔｉ…     

相关场的投影理论

本文应用Radon变换导出了二维相关场的一维投影积分方程.并通过构造积分方程的投影算子实现了二维相关场向一维相关场的变换.并对—维投影相关特性进行了理论分析.     

基于方向-频率分解的旋转不变性纹理分类

提出了一种用于纹理分类的旋转不变性特征提取的新算法.该算法是将一定大小的图像进行二维傅里叶变换;其次在变换后的图像中央选择一个圆盘区域,并在方向[0°,180°]内进行等间隔角度频率抽样,实现方向分解,使用一组复Morlet小波对每个方向上的映射切片进行小波变换,从而实现多通道频率分解;在各个频率通道中计算均值和方差作为特征,并利用线性回归模型计算频率通道之间的关系特征;将特征沿方向进行一维傅里叶变换并取其幅值,从而得到旋转不变性特征.实验结果表明所提取的特征具有较好的旋转不变性,与其它算法相比具有更好的分类性能,并且对无旋转纹理分类也能产生较好的分类结果.     

一种改进型视角投影图像彩色计算全息术的三维重构算法

本文提出了一种利用三维景物的二维视角投影图像合成计算全息图,并重构出彩色再现三维影像的方法.该方法基于利用视角投影图像获取景物的三维傅里叶频谱的理论,采用电荷耦合器件记录三维景物在白光照明条件下横、纵两正交方向的一系列视角投影图像,并利用这些视角投影图像合成计算全息图,从而重构出三维再现像.通过采用在频谱面上的容余采样方法,提高了图像频谱信息的利用率,通过实验论证,证明了该方法的可行性.利用该方法使得视角投影图像的记录过程更加简单,节省了采样时间,提高了程序运行速度|能够在利用同等数量的视角投影图像的条件下,提高合成全息图的质量,使得重构的彩色再现三维影像更加清晰.     

Quantum Multi-Image Encryption Based on Iteration Arnold Transform with Parameters and Image Correlation Decomposition

A novel quantum multi-image encryption algorithm based on iteration Arnold transform with parameters and image correlation decomposition is proposed, and a quantum realization of the iteration Arnold transform with parameters is designed. The corresponding low frequency images are obtained by performing 2-D discrete wavelet transform on each image respectively, and then the corresponding low frequency images are spliced randomly to one image. The new image is scrambled by the iteration Arnold transform with parameters, and the gray-level information of the scrambled image is encoded by quantum image correlation decomposition. For the encryption algorithm, the keys are iterative times, added parameters, classical binary and orthonormal basis states. The key space, the security and the computational complexity are analyzed, and all of the analyses show that the proposed encryption algorithm could encrypt multiple images simultaneously with lower computational complexity compared with its classical counterparts.     

Computer-Generated Holograms Based on the Three-Dimensional Fourier Spectrum

A method to synthesize a computer-generated hologram (CGH) of real-existing objects from projection images is proposed. Different from other similar methods, our method enables the synthesis of a CGH with only one-dimensional (1-D) mechanical scanning of the objects. Our method is connected with the three-dimensional (3-D) Fourier spectrum of the objects by the 3-D central slice theorem (CST). Two efficient recording techniques for projection images and numerical experiments to verify our principle is discussed. A comparison between the two techniques is also presented from the viewpoint of diffraction efficiency. © 2005 The Optical Society of Japan     

Period correction method for binary fringe defocused projection

Binary fringe defocused projection can resolve the problem caused by the nonlinear gamma of the projector. Owing to the intersecting axis measurement system, the broadening of the fringe period on the reference plane can cause measurement errors. Non-uniform periodical binary fringe defocused projection is utilized to overcome this problem. After appropriate defocused projection of non-uniform periodical binary fringe, uniform periodical sinusoidal fringe can be obtained on the reference plane. This method can prevent the nonlinear gamma effect and broadening of the fringe period, and filter high harmonics and high-frequency noise. Three-dimensional (3-D) shape measurement experiments of standard flat are performed with four-step phase-shift method. Experimental results demonstrate that the proposed method exhibits high measurement precision. Highly accurate 3-D measurements of large objects can also be performed with the proposed method.     

Three-Dimensional Large Screen Display Using Polymer-Dispersed Liquid-Crystal Light Valves and a Schlieren Optical System: Proposal and Basic Experiments

A novel three-dimensional (3-D) projection display used with polarized eyeglasses is proposed. It consists of polymer-dispersed liquid crystal-light valves that modulate the illuminated light based on light scattering, a polarization beam splitter, and a Schlieren projection system. The features of the proposed display include a 3-D image display with a single projector, half size and half power consumption compared with a conventional 3-D projector with polarized glasses. Measured electro-optic characteristics of a polymer-dispersed liquid-crystal cell inserted between crossed polarizers suggests that the proposed display achieves small cross talk and high-extinction ratio.     

用于指纹／字符识别的电子投影相关技术

本文应用二维图像场的一维投影相关理论,有效而迅速地实现了指纹及字符场的识别,并得到了较满意的识别结果.     

复杂刚性运动物体的高精度三维测量算法

基于相移法的三维形貌重建精度高,对环境噪声和阴影等不敏感,但由于多幅条纹解相位,难以应用于动态物体的三维测量中,为此,提出了一种新的算法.基于Harris算法提取刚性运动棋盘格的角点,确定相邻两帧采集条纹图像之间的像素偏差并校正采集条纹图像;根据投影仪和摄像机的标定参数建立投影图像和采集图像之间的空间变换矩阵,并根据变...     

小波方法分析小流量喷气影响两级低速轴流压气机失速特性的动态效果研究

本文用连续小波变换和二维图像处理方法对三个不同角度的喷气试验数据进行了分析．通过对比分析，连续小波变换发现失速前转速频率成分的能量在两级动叶前的信号中出现波动现象，而这种波动的强度与喷射角有关；二维小波图像分析发现，压力场中相对于叶片通道而言的大尺度奇异性均在失速前呈波动性增强，这种波动以接近转子转速沿压气机周向旋转，喷射角度对失速前的奇异性及其波动程度有明显影响．一维小波方法和二维方法的结果相互对应，但二维方法更加具有优势．     

二维轴对称图象高阶胡氏不变矩快速算法的改进

提出了基于二维轴对称性目标图象矩快速算法的实用化改进方案.算法实用化改进是通过两个命题的证明实现的.一是通过轴对称目标图象质心在对称轴上的命题证明,实现了将二维轴对称图象矩快速算法用于轴对称性图象中心矩的计算;二是通过目标图象以坐标系中任意两点为圆心,旋转相同角度,结果图象具有平移性的命题证明,实现了算法用于对旋转不同角度的对称目标图象的中心矩的计算改进.在此基础上,完成了二维轴对称目标图象高阶胡氏不变矩实用化快速算法的实现.实验证明,该算法具有较好的实时性能,且具有较小的引入误差.     

Incoherent holographic imaging through thin turbulent media

We introduce and experimentally demonstrate a new method of computing Fresnel holograms of a real-existing, three-dimensional (3-D) scene hidden behind a thin turbulent medium, and illuminated by incoherent white light. From each perspective point of view, we acquire multiple noisy speckled images through the medium vibrated mechanically. These images are fused together to yield a smoothed perspective projection of the hidden 3-D scene. All smoothed projections are processed to yield a Fresnel hologram of the hidden scene. The 3-D image reconstructed from the hologram is further improved by digital blind deconvolution of each of the 3-D image slices with its own estimated impulse response.     

光电成像探测中目标方位测量方法

为了在光电成像探测目标的同时给出目标相对探测器的方位信息,探讨了一种基于单目视觉的运动目标方位测量方法。根据投射成像原理,推算出二维目标像点坐标与目标的空间三维位置之间的映射关系,建立了单目视觉测量目标方位的数学模型。结合帧间差分法和KLT方法检测静止背景下和变化背景下的运动目标,并对目标特征点进行亚像素定位。试验和仿真结果分析表明,该方法能够有效提取目标的特征点,可对目标较为准确地定位,并通过单目视觉测量出目标方位信息,误差控制在8%的范围内。     

The future of 3-D range image measurement using binary-encoded pattern projection

3-D range image measurement using image processing is well established in a variety of applications as a reliable industrial measurement technique. Various triangulation-based techniques have been developed, using temporal or spatial binary encoding to uniquely identify each measured point. These are reviewed, and a new adaptive technique using a binary encoded sequence, combining spatial and temporal variation, is described.     

3-D shape measurement based on complementary Gray-code light

A combination of phase-shift with Gray-code light projection into a three-dimensional (3-D) measurement system has been exploited to digitalize 3-D shape information of a tested object, even with a discontinuous surface. Unfortunately, the phase unwrapping will fall into an error, when an improper value of Gray coding is caused by mistake at the partial boundary of two adjacent binary words. To this end, a new complementary Gray-code method is proposed in this paper as well as the corresponding phase-unwrapping method. This problem of phase unwrapping could be cleverly solved by projecting an additional Gray-code pattern to extend this code and using the different and complementary boundary locations of the traditional and additional codes. The results of computer simulation and experiment confirm that this proposed method based on complementary Gray-code can reliably reconstruct the nature phase distribution of the tested object with only one extra fringe pattern.     

位错点阵投影的三维数字成像

提出一种基于位错点阵投影的三维数字成像方法。该方法以二维点阵结构光顺序照明参考平面和被测物体表面,通过分析分别投影在参考平面和物体上的点阵的成像过程,建立投射在参考平面和被测物体表面上的点阵中同一点在成像面上横向位置错动与被测物体深度之间的数学模型。依据此模型,通过计算相对应点在成像平面上的横向位移。可计算出物体的深度图像。二维点阵可以在垂直于光轴的平面内做横向和纵向的两维移动,以填补由于离散而丢失的其他空间点的深度信息,获得高空间分辨率。实验验证了该方法的可行性,此方法对于拓扑复杂的表面和大梯度表面有较强的适应性。     

高速投影云纹技术的计算机数据处理

本文采用数字图象处理技术,最小二乘法曲线拟合原理和计算机图形学原理,对铝板在点爆驱动下的高速云纹图象(拍摄频率为5×10⁴ps)进行计算机数据处理, 重建了铝板在不同时刻的三维形状, 并使高速投影云纹技术的测量精度达到了±5%。     

基于点阵编码的三维主动视觉标定

提出一种基于位错点阵编码三维成像系统的标定方法.首先在物空间建立三维数据基准,然后通过基准传递的概念标定摄像子系统,再通过建立摄像子系统坐标系与投影子系统坐标系之间的约束关系,将标定后的摄像子系统的准确度传递到投影子系统坐标系.摄像子系统坐标系与投影子系统都具有标定的准确度之后,可以根据位错点阵编码三维成像技术的解码算法获得深度图像空间坐标的计算值,然后将其与物体空间的三维标定数据基准进行比较,建立目标函数为误差平方和最小的非线性优化方程.通过迭代求解这个优化方程,最终获得三维系统的结构参量.实验结果表明,经过三维标定的位错点阵编码三维成像系统,对300×300×80 mm3的测量体积内,可以获得X方向的标准差为0.29mm和Y方向的标准差为0.24mm,Z方向的标准差为0.29mm的测量准确度.     

物体内部三维位移场分析的数字图像相关方法

提出了物体内部三维位移场的数字图像相关分析方法，对物体变形前后，或连续变形的两个相邻状态的内部三维结构的数字图像，通过相关运算获得三维位移场.文中给出了三维相关法的体搜索窗口、相关函数及亚像素运算的相关系数拟合函数.数字模拟结果证明了三维相关法的正确性及可靠性.位移计算精度为0.02像素. 关键词： 数字图像相关 三维相关 亚像素