A novel probabilistic method for robust parametric identification and outlier detection

Outliers are one of the main concerns in statistics. Parametric identification results of ordinary least squares are sensitive to outliers. Many robust estimators have been proposed to overcome this problem but there are still some drawbacks in existing methods. In this paper, a novel probabilistic method is proposed for robust parametric identification and outlier detection in linear regression problems. The crux of this method is to calculate the probability of outlier, which quantifies how probable it is that a data point is an outlier. There are several appealing features of the proposed method. First, not only the optimal values of the parameters and residuals but also the associated uncertainties are taken into account for outlier detection. Second, the size of the dataset is incorporated because it is one of the key variables to determine the probability of obtaining a large-residual data point. Third, the proposed method requires no information on the outlier distribution model. Fourth, the proposed approach provides the probability of outlier. In the illustrative examples, the proposed method is compared with three well-known methods. It turns out that the proposed method is substantially superior and it is capable of robust parametric identification and outlier detection even for very challenging situations.     

基于Daubechies小波的图像边缘检测技术

利用Daubechies正交小波变换的性质,通过Mallat多尺度分析方法对图像进行小波变换,把图像分解成低频轮廓,水平高频、垂直高频和斜线高频四个部分。针对图像边缘主要集中在高频部分,该文先保持小波变换后的高频小波系数,同时对低频小波系数进行再次小波变换,提取出次高频信号的边缘信息。最后对保留下来的高频小波系数和次高频小波系数进行逆变换获取最大边缘信息。     

A one-class peeling method for multivariate outlier detection with applications in phase I SPC

In phase I of statistical process control (SPC), control charts are often used as outlier detection methods to assess process stability. Many of these methods require estimation of the covariance matrix, are computationally infeasible, or have not been studied when the dimension of the data, $p$, is large. We propose the one-class peeling (OCP) method, a flexible framework that combines statistical and machine learning methods to detect multiple outliers in multivariate data. The OCP method can be applied to phase I of SPC, does not require covariance estimation, and is well suited to high-dimensional data sets with a high percentage of outliers. Our empirical evaluation suggests that the OCP method performs well in high dimensions and is computationally more efficient and robust than existing methodologies. We motivate and illustrate the use of the OCP method in a phase I SPC application on a $N=354$, $p=1917$ dimensional data set containing Wikipedia search results for National Football League (NFL) players, teams, coaches, and managers. The example data set and R functions, OCP.R and OCPLimit.R, to compute the respective OCP distances and thresholds are available in the supplementary materials.     

新的梯度边缘检测方法

提出了一种新的基于梯度算子法检测灰度图像边缘的方法。该方法在3×3的邻域窗内,分别用梯度法对窗内的水平方向、垂直方向和正、负45°角方向上的边缘线进行检测,找到梯度最大的边缘线,从而组成最终的边缘图像。实验表明:该方法简单,边缘检测图像效果好。在此基础上,又提出了增强边缘检测方法和简化边缘检测方法。这两种对基于梯度算子法的改进方法分别适用于检测小的边缘梯度和需要快速检测边缘这两种情况。     

采用阈值的快速边缘检测算法

针对原有模糊边缘检测算法计算复杂、适应性差的缺点,提出一种新颖的快速边缘检测算法。该算法将图像映射为一个模糊特征面,通过图像自适应阈值来确定模糊增强的渡越点。在渡越点上对图像进行非线性增强。算法将“Min”与“Max”算子结合起来提取边缘,避免了使用单一算子检测边缘强度较弱的缺点。该方法避开了原算法中复杂的指数运算,运算量减少了40%左右;渡越点的自适应性提高了算法的适应性。仿真表明,该算法检测的边缘细、连贯,并且算法抗噪能力强。     

基于多元角域指标离群检测的风电齿轮箱故障预警方法

针对风速和载荷随机波动造成的早期故障特征难以提取和量化的特点,提出基于多元角域指标离群检测的风电齿轮箱故障预警方法。首先利用阶比重采样技术将非平稳的时域振动信号转化为具有平稳或准平稳特性的角域信号,提取角域无量纲指标反映风电齿轮箱早期故障趋势;其次利用多元相关度测量建立风电齿轮箱早期故障识别模型;最后采用多元离群检测方法实现风电齿轮箱早期故障预警。实验表明,该方法能够实现较为准确地的风电齿轮箱早期故障预警,具有较好的工程实际意义。     

改进的直线拟合线阵CCD图像边缘检测方法

为了检测经过光学系统成像所得图像的边缘,在直线拟合方法的基础上提出了一种新的边缘检测方法。该方法综合考虑边缘过渡区间所有点的信息,利用直线拟合方法求出一组可能的边缘值,再选择它们的中值作为最终的边缘值。基于该方法的采集系统由线阵 CCD 采集边缘的灰度图像,使用单片机来处理数据,并计算出边缘的位置。实验结果表明,这种改进方法的重复性误差为 0.89 个像素,而且能够在 0.039s 内完成一次边缘检测计算。     

基于变差函数的噪声图像的多尺度边缘检测

基于区域化变量理论,针对受噪声污染的图像,本文提出了一种基于变差函数的多尺度边缘检测新方法.该方法根据图像各个不同区域的数据的不同方向的变差函数值来判断该区域是否存在边缘以及边缘的方向性,然后根据该区域边缘的方向性,在水平和垂直方向分别进行不同尺度的小波变换,进而达到在确保边缘定位准确的同时,尽最大可能去除由于噪声以及图像灰度不均匀产生的伪边缘点.仿真实验表明,本文算法在对受高斯白噪声污染较严重的图像进行边缘检测时能有效的去除噪声对图像边缘检测的影响,从而证明了该方法的可行性、有效性.     

图像的模糊边缘检测算法

常用的微分边缘检测算法往往无法设立合适的阈值将影像中梯度较小的模糊边缘检测出来.针对这一点,疚奶岢隽肆街纸饩龅姆椒?将图像方差标准化,拉大模糊边缘处的梯度值,或者通过设置sigmoid函数,将像素点所在区域的信息传递到梯度值中去,对梯度值进行调整,这样就能够设定合适的阈值,有效地将模糊边缘提取出来.本文将这两种算法和常用的一些微分边缘检测算法比如Sobel、LOG算法进行了比较.试验表明,这两种方法都能够有效地提取出模糊边缘.     

基于边缘矢量的镰刀弯在线检测方法研究

针对热连轧车间里带钢的镰刀弯检测问题,提出一种基于机器视觉的镰刀弯在线检测方法.首先,对预处理后的图像采用图像二值化处理提取目标,再使用形态学运算消除目标上的杂质以及不连通区域.然后在图像中以图像的中心点像素作为原点建立坐标系,提取中间坯两侧边缘点,接着获取相邻点之间的向量,并进行单位绝对值化,计算相邻单位向量之间的关...     

Parameter optimal determination for canny edge detection

Abstract

The Canny edge detection algorithm contains a number of adjustable parameters, which can affect the computation time and effectiveness of the algorithm. To overcome the shortages, this paper proposes a new way to determine the adjustable parameters and constructs a modified Canny edge detection algorithm. In the algorithm, an image is firstly smoothed by an adaptive filter that is selected based on the properties of the image, instead of a fixed sized Gaussian filter, and then, the high and low thresholds for the gradient magnitude image are determined based on maximum cross-entropy between inter-classes and Bayesian judgment theory, without any manual operation; finally, if it needs, the object closing procedure is carried out. To test and evaluate the algorithm, a number of different images are tested and analysed, and the test results are discussed. The experiments show that the studied algorithm can achieve the better edge detection results in most of the cases, and it is also useful for object boundary closing as a pre-segmentation step.     

一种改进多尺度形态边缘检测算法

提出一种改进的多尺度形态边缘检测算法。用不同尺度大小的结构元素分别检测出图像的不同尺寸的边缘信息,运用本文提出的基于直方图配准的自适应方法对所获不同尺寸的图像边缘进行融合,获得理想的图像边缘,有效地消除了噪声。实验结果表明当图像中加入了强度超过20％的椒盐噪声后,检测到的边缘仍然非常清晰。     

Progressive edge detection compression for fingerprint images

In this article, a progressive fingerprint image compression (for storage or transmission) that uses an edge detection scheme is developed. First, the image is decomposed into two components: the primary component, which contains the edges, and the secondary component, which contains the textures and the features. Then, a general grasp for the image is reconstructed in the first stage at bit rates of 0.0223 and 0.0245 bpp for the tested fingerprints images (samples 1 and 2), respectively. The quality of the reconstructed images is competitive with the 0.75‐bpp target bit set by the FBI standard. Also, the compression ratio and the image quality of this algorithm are competitive with other methods reported in the literature. The compression ratio for our algorithm is about 45:1 (0.180 bpp). © 2003 Wiley Periodicals, Inc. Int J Imaging Syst Technol 12, 211–216, 2002; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ima.10025     

一种改进的Snake眼睛边缘检测算法

提出了一种应用于彩色图像的改进的Snake眼睛边缘检测算法.该算法首先将输入图像从RGB空间转换到YCbCr和YCb'Cr'空间,依据肤色点在Cb'Cr'空间中的聚类性提取人脸候选区域;然后通过在人脸候选区域上生成眼睛图(EyeMap)提取眼睛候选区域并计算候选区域Cr2图像的距离势能力场.采用两段首尾相连的抛物线进行初始化并从外部对Cr2图像的距离势能力场进行活动轮廓模型(Snake)匹配,在匹配过程中不断进行修正使检测结果更接近真实边缘.实验结果证明,该算法能精确地检测出眼睛的边缘并节省匹配时间.     

模式密度方法在结构异常检测中的应用研究

结构损伤发生后,结构的动力特性将随之改变,因此可通过检测动力特性的异常来判断结构的健康状况.研究了基于模式密度的异常检测在损伤诊断中的应用,该方法具有无需建立训练模型,能够支持模式动态更新等优点.以ASCE学会提出的基准结构为对象,采用基于相空间重构的奇异值分解技术提取适当个数的奇异值构造特征模式集,应用模式密度方法检测结构异常状况.仿真和试验结果表明,该方法能有效地检测损伤的发生.     

综合边缘检测和区域生长的红外图像分割方法

针对红外图像的特点,提出了一种综合应用边缘检测和区域生长方法的图像分割方法。其思路为:先对图像进行边缘提取,得到边缘像素点集;然后利用该点集的平均灰度和目标区域的连通性作为生长判决条件,采用区域生长法实现图像分割。仿真结果表明,该方法能快速准确有效地实现红外图像分割,避免了单独使用边缘提取或区域生长法进行图像分割时的典型分割错误。     

剔除异常数据的稳健性处理方法

由于以往的剔除异常数据的方法,包括国家标准GB8056-87和GB6380-86中所给出的判别异常数据的方法,多利用残差和标准差且基于假定某种典型概率分布,故存在其局限性.再者,在初始判别时就可能受异常数据的影响,尤其在小样本下会有误判.经作者各种各样的仿真试验,提出一些剔除异常数据的稳健性通用处理方法,以期适于在不同概率分布的小样本下应用.     

一种提取图像目标边缘的新方法

提出一种基于阈值分割技术和形态学相结合的图像目标边缘提取方法,用Top-hat变换后的图像与原始图像相加再减去Bottom-hat变换后的图像以得到最大对比度的图像,继而进行自动图像阈值分割,再用圆形模板执行形态学闭合操作提取边缘,完全去除内部不感兴趣的细节,并保持边缘的连贯性。仿真结果表明,该方法能很好地提取目标边缘,而实际计算量只有传统方法的44.6%～70.2%,且具有较好的抗噪声能力。     

多阈值选取与边缘连接的边缘检测算法

论文针对Kirsch边缘检测算子的缺点,提出一种多阈值选取与改进的边缘连接算法相结合的方法,对图像,尤其是医学图像进行边缘检测。该方法首先用高斯滤波器对原始图像去噪,使用传统Kirsch算子计算梯度,然后设定低阈值和自适应阈值来保护图像弱边缘和提取真正边缘,最后通过改进的边缘连接算法连接边缘。实验结果表明,论文的方法具有保持弱边缘、连接性好的边缘检测效果。