基于局部分形维数的遥感图像分割

本文针对遥感图像的分形特点,采用局部分形维数进行图像分割.利用金字塔模型替代了局部分形计算中的窗口模板,从而对局部三角棱柱面法进行了改进,减少了图像局部分形维数的运算量.并设计了一种新的递归计算流程,有效地降低了内存存储量.利用此法对遥感图像进行分割,结果表明:改进后的三角棱柱面法提高了基于遥感图像纹理特征的图像分割方法的计算速度,同时保留了三角棱柱面法对于分形维数计算的准确性.     

分形维数在判断海洋锋强度中的应用

提出一种用计算维数值预报海洋锋强度的方法。给出了用分形维数描述海洋锋强度的理论依据,介绍了分形维数的计算方法及现有的海洋锋强度的判断方法。应用网格法,以HOOM海洋锋模型和我国东海黑潮锋区的三维声速场数据为例,计算了三维声速场中不同深度、时间的声速等值线的分形维数,得出HOOM模型海洋锋声速场分维值在1.07和1.60之间,我国东海黑潮锋区分维值在1.04和1.30之间,无锋时声速等值线分维接近于1;分形维数可以表征海洋锋强度,维数值随锋强度的变化而变化的结论。     

分形图像分析及其在纳米ZnO中的应用

研究了分形图像的分析过程,利用MATLAB软件完成了分形维数计算程序的设计.并提取了纳米ZnO透射电镜像的分形图像,利用分形维数计算程序计算其分形维数,证明这种图像数字化方法能对纳米ZnO的形态进行评判,从而实现了定量描述纳米ZnO的形态.     

基于数学形态学的分形维数计算及在轴承故障诊断中的应用

滚动轴承故障信号是一种典型的非线性信号,分形几何为描述轴承故障信号的特性提供了一个有力的分析工具。基于数学形态学的分形维数是在Minkowski-Boulingand维数基础上拓展的一种采用形态学操作计算分形维数的新方法。本文较详细的阐述了基于数学形态学的分维数计算方法,对比分析了与传统计盒维数方法的区别与联系,并对实际的滚动轴承正常、滚动体故障、内圈故障和外圈故障信号进行了分析,结果表明,基于数学形态学的分维数计算方法具有计算速度快,估计准确稳定的特点,为准确判断滚动轴承故障状态提供了一种快速有效的新方法。     

基于离散分数布朗随机场的水下图像目标检测

采用传统图像检测方法存在目标区域定位不准确、目标细节信息丢、目标形状变形等问题,本文提出一种基于离散分数布朗随机场模型的水下图像目标检测方法.该方法根据分形理论和水下图像的特点,以图像中每个像素点为中心取窗口,计算在该窗口内的分形维数均值,将该均值作为中心像素的分形特征,然后根据分形维数分布图确定分割阈值,从而实现对水下图像分割,并且通过将目标表面不同尺度下的灰度差分平均值进行归一化处理,减少了用于表示不同尺度下的平均绝对值灰度差分的数据,从而提高算法检测效率.实验结果表明,该方法对水下成像条件具有一定鲁棒性,是一种有效的水下图像目标检测方法.     

爆破振动信号时频特征的三维分形特性研究

为进一步研究爆破振动信号能量特征,基于小波包变换方法对爆破振动信号时频能量特征进行分析。在二维矩形盒维数模型的基础上建立了计算三维曲面分形盒维数模型,并计算得到实测爆破振动信号时频能量谱的三维分形盒维数D3d=2.148 8。该分形维数满足空间分形条件2D3d3,验证了爆破振动信号时频能量谱具有三维分形特征。经研究三维分形曲面盒维数D3d与其剖面二维分形盒维数D2d之间的数值关系,表明计算空间分形曲面分维数的常规假设D3d=D2d+1并非严格成立。研究表明,所提出的长方体覆盖空间曲面的分形维数计算方法可行、有效,为进一步研究爆破振动信号特征、控制爆破地震效应提供了新的研究思路。     

基于增强分形特征的人造目标检测

自然背景和人造目标对于某些分形特征存在着一定的本质差别,这些差别为目标检测提供一套全新的方法。研究表明,对于自然背景中嵌入少量人造目标这类简单情况,采用单一的分形特征就能取得较好的检测效果。对于复杂情况下的检测,本文采用Sarkar和Chaudhuri的差分计盒（DBC）分数维法计算图像的分维数,并用概率松弛法（PRIA）对分维值进行特征增强。仿真实验表明,该方法具有较好的检测效果。     

材料断裂形貌分维的测量及计算机模拟

材料断裂形貌如断口表面、断裂裂等均被认为具有分形结构而被称为实际分形体,论述了这类实际分形体维数的实验测量、计算方法和分维测量、计算中出现的分维不性问题,指出生分维不确定性的原因,并且通过计算机模拟演示了一些实际测量因素对测量和计算实际分形体维数的影响。     

分形维数在原棉异纤图像处理中的应用

图像的分形维数是图像的重要定量特征, 可广泛用于图像分析当中.本文采用计盒法计算出图像的分形维数, 根据分形维数的差异判断原棉中是否有异纤, 结合分形维数的 Ostu 改进算法对有异纤原棉图像进行分割.     

基于分形理论的岩石冲击破坏分形分析

利用大直径分离式霍普金森压杆试验系统,对绢云母石英片岩和砂岩进行冲击压缩试验, 采用不同等级标准筛对岩石冲击破碎后试块进行筛分统计,运用分形几何理论,计算出冲击荷载作用下两种岩石破碎块度分布的分形维数,研究冲击速度对块度分维影响,分析两种岩石动态抗压强度随块度分维的变化关系。试验结果表明,绢云母石英片岩块度分维大部分集中在1.9~2.4之间,砂岩集中在2.5~3.0之间;两种岩石分形维数随冲击速度的升高呈上升趋势,近似线性正比关系;绢云母石英片岩动态抗压强度与块度分维无明显函数关系,砂岩动态抗压强度随块度分维的增大呈增加趋势。采用分形维数对岩石试件在冲击破碎过程中动态抗压强度的变化进行定量描述,为探索岩石动态破碎分形特征与冲击力学性能之间的内在规律,开辟新的研究途径。     

大气颗粒物显微轮廓分形与源解析

为监测大气颗粒物的来源,根据静电吸附原理设计大气颗粒收集装置采集大气颗粒物,通过光学显微放大技术和图像分析方法,得到源颗粒的轮廓分形维数值,结合采样点周边的实际情况,分析和对比各源颗粒的轮廓分形维数值,确定颗粒部分来源;并采用偏光显微镜观察各源颗粒的偏光特征,协同确定大气颗粒的具体来源。结果表明,大气颗粒具有分形特征,其很多特性均与其分形特征有关;采用大气颗粒的轮廓曲线分形维数值作为降尘颗粒来源的辨析参数是可行的,轮廓曲线的分形维数值能很好地表达颗粒的结构形态,特定颗粒源的分形维数值具有一定的范围;将轮廓曲线的分形维数值与各源颗粒的偏光特征相结合,对颗粒来源的辨识准确率在90%以上。     

复合评定法在三维粗糙表面评定中的应用

提出一种将双树复小波与分形理论相结合的复合评定方法。即利用双树复小波近似的平移不变性和方 向性好等优点,将信号分解成更为细腻的低频和高频信号,同时根据多尺度下信号分形维数的不变性,利用图像灰 度值自相关求得高低频信号的分形维数,通过计算高低频信号分形维数之间的差值（即分形维数距)来确定双树复 小波的分解尺度。仿真结果表明复合评定法可以较好地提取基准面,且分形维数距确定分解层数的准确性通过均 方根*Sq值得到了验证;两个实例均说明,复合评定法可以很好地提取具有分形特征的纳米级三维粗糙表面基准或 波纹度,为实际工程表面评定提供了可靠的理论基础。     

高温蠕变中的金相分维变化

对经高温持久强度试验的试样，沿其轴线拍摄一系列金相照片，用周积法测算出各自的维数，得出维数沿轴向的分布规律。从中发现，蠕变程度越严重的区域，其分形维数越小，在断裂的地方达到最小。在工程上，可以利用其分维最小值与平台区的分维来判断炉管的安全性。     

Multi-receivers and sparse-pixel pseudo-thermal light source for compressive ghost imaging against turbulence

For remote sensing with computational ghost imaging, the atmospheric turbulence will influence much on the performance since the transmitted patterns from the pseudo-thermal source will be distorted. As the distribution of refractive index on the propagation path varies with time, reducing the time consumption of data acquisition will be an effective way for mitigating the influence. We propose using fewer pixels of spatial light modulator to increase the modulation rate of the pseudo-thermal source, with the positions of pixels being randomly distributed. Multi-bucket detectors are employed for further reducing of the data acquisition time. The sparse recovery algorithm from compressed sensing is employed to enhance the quality of image. The performance of this system is demonstrated by comprehensive numerical analyses.     

Fractal dimension of metallic fracture surface

In this study, a complete method of determination of the fractal dimension for fracture surfaces of ferrous alloys has been proposed. This dimension is determined for the vertical profile obtained by the profile technique cross-section. The image of the profile, seen through the microscope coupled with a camera, is recorded in a computer, where numerical processing is performed. For calculation of the same fractal dimension, the fd3 program has been used, which is available through the Internet. The essential element of the method is optimisation concerning microscopic magnification (scale of a length), resolution of the recorded image and selection of the grey level threshold at binarization. The tests for the stability of discretization, which enable minimization of the error of the measurement, have also been carried out. These tests consist in checking the difference in fractal dimensions for the same profile obtained in two different methods of contouring as well as the difference between capacitive, informative and correlative dimensions. In both cases, too big difference suggests that the determined dimension is not reliable. This method allows determination of the fractal dimension with an absolute accuracy of 0.05 in non-dimensional units. The method has been employed in many studies. In this paper the following tests have been presented: a “fractal map” of the fracture surface was made, an influence of the mechanical notch radius in a compact specimen on the fractal dimension of the fracture surface, an influence of the distortion rate on the fractal dimension, an effect of fatigue crack propagation rate on the fractal dimension and influence of the stress-intensity factor on the fractal dimension of the fracture surface. The following materials were examined: Armco iron, P355N steel and 41Cr4 steel in different states after the heat treatment. The measurements have been made for the specimens of the compact type. There was considered an influence of location of the place of measurement on the fractal dimension being determined. The dimension was determined on the profiles lying longwise and crosswise the crack propagation direction. It has been found that the fractal dimension of the fracture surface does not depend on a place of measurement. This suggests, among other things, that a distinction between the places, which were created under conditions of the plane stress, and the places, which were created under conditions of the plane strain state, cannot be made with the help of the fractal dimension. When testing an influence of the radius of the mechanical tip notch on the fractal dimension of a fracture surface, this dimension was determined in the places located at different distances from the tip of the mechanical notch. With respect to the radii up to 1.0 mm, no significant differences in fractal dimensions have been found. The fractal dimensions of the fracture surface for all examined materials were practically the same and they ranged from 2.02 to 2.10. However in some ranges of da/dN rate the dimension was changing inversely proportional to da/dN. Obtained results confirm that fractal dimension do not depend on the investigated material.     

Statistical artefacts in the determination of the fractal dimension by the slit island method

This paper comments upon some statistical aspects of the slit island method which is widely used to calculate the fractal dimension of fractured surfaces or of materials’ features like grain geometry. If a noise is introduced when measuring areas and perimeters of the islands (experimental errors), it is shown that errors are made in the calculation of the fractal dimension and more than a false analytical relation between a physical process parameter and the fractal dimension can be found. Moreover, positive or negative correlation with the same physical process parameter can be obtained whether the regression is performed by plotting the variation of the noisy area versus the noisy perimeter of the considered islands or vice versa. Monte-Carlo simulations confirm the analytical relations obtained under statistical considerations.     

海洋遥感彩色图像中的海洋数据读取技术

基于对海洋遥感彩色图像的像素点数据矩阵的理解,提出了一种基于MATLAB软件的海洋遥感数据读取技术.依据这项技术,用户可以从卫星遥感获取的各种海洋要素彩色图像中读取对应于各个像素点的海洋要素数据.在MATLAB软件中,图像是由像素点构成的,全部图像信息均以矩阵的形式存储和运算.一幅图像可能包含一个数据矩阵和一个颜色映射表矩阵.用户可使用MATLAB软件读入图像,并通过所读取图像中包含的数据矩阵和颜色映射表矩阵,将图像区域中的像素点的颜色与色标中的颜色进行比较来获得像素点所表示的海洋要素数据.以一幅东中国海海域的海表面温度(SST)彩色图像实例说明了如何应用该技术来提高遥感彩色图像信息的读取能力.     

超细炸药高氯酸·[四氨·双(5-硝基四唑)]合钴(Ⅲ)的粒度分维特征

测试了高氯酸·[四氨·双(5-硝基四唑)]合钴(Ⅲ)原样及其超细样品的粒度分布。结果表明,超细样品和原样均呈双峰/多峰分布,且其粒度分布具有分维特征。原样的分维数较低为1.80,3个细化样的分维数均较高,为2.32～2.42。分析认为,研磨时间从0延长至60min,对应样品粒度分布分维数增大,符合通常的粒度分布规律。研磨时间从60min延长至150min期间,对应样品中亚微米和纳米粒子含量较高,团聚较严重,使粒度分布发生异常,粒度分布分维数计算中的相关系数有所减小,分维数计算值不再增加。分维数可在一定程度上反映超细粒子团聚行为的严重程度。     

A Weighted Threshold Secret Sharing Scheme for Remote Sensing Images Based on Chinese Remainder Theorem

The recent advances in remote sensing and computer techniques give birth to the explosive growth of remote sensing images. The emergence of cloud storage has brought new opportunities for storage and management of massive remote sensing images with its large storage space, cost savings. However, the openness of cloud brings challenges for image data security. In this paper, we propose a weighted image sharing scheme to ensure the security of remote sensing in cloud environment, which takes the weights of participants (i.e., cloud service providers) into consideration. An extended Mignotte sequence is constructed according to the weights of participants, and we can generate image shadow shares based on the hash value which can be obtained from gray value of remote sensing images. Then we store the shadows in every cloud service provider, respectively. At last, we restore the remote sensing image based on the Chinese Remainder Theorem. Experimental results show the proposed scheme can effectively realize the secure storage of remote sensing images in the cloud. The experiment also shows that no matter weight values, each service providers only needs to save one share, which simplifies the management and usage, it also reduces the transmission of secret information, strengthens the security and practicality of this scheme.