分形理论在近表面缺陷超声A扫检测中的应用

针对超声相控阵无损检测(NDT)中近表面缺陷识别困难的问题,提出一种基于分形理论的近表面缺陷智能识别方法。运用基于线性插值的盒计数维数算法,计算140组超声A扫(A-Scan)信号的盒计数维数,并运用统计的方法详细分析其分布情况。实验结果表明超声A-scan信号具有分形特性,分形理论可应用于A-Scan信号分析;而且有无缺陷信号的盒维数分布区间差异明显,盒维数可作为A-Scan信号的特征识别近表面缺陷。在超声相控阵自动化检测中,运用分形理论能提高近表面缺陷的检出率,减少人为因素引起的漏检。     

基于分形维数的磁痕图像缺陷检测

针对荧光磁粉无损探伤自动检测中存在的磁痕识别率低、虚警率高、抗干扰能力弱和易受背景影响的问题,提出一种基于分形维数的磁痕图像缺陷在线检测方法。采用差分计盒维数算法求解得到磁痕图像的分形维数,并分析工件磁痕缺陷分形维数特征。实际运用中,实时采集工件图像,利用特征进行图像处理,判断工件是否存在缺陷。经过实验验证,采用基于分形维数特征的磁痕图像缺陷检测方法,能准确识别工件缺陷,磁痕识别率为80%,与常用检测磁痕方法相比,虚警率降低5%,受背景影响小,有较强的适用性。     

彩色图像分形维数的计算方法

分形维数是描述图像复杂性的一种重要测度,广泛应用于图像特征提取及图像分类、分割和检索等方面。多种黑白图像和灰度图像的分形维数计算方法已被提出,但其中很少有适用于彩色图像的分形维数计算方法。把计算灰度图像分形维数的差分盒维法扩展到欧氏五维空间,提出了一种简单且易实现的计算彩色图像分形维数的方法。实验结果表明,提出的方法能够捕捉到彩色图像纹理的复杂性,在识别彩色图像粗糙度变化和计算精度方面优于其它算法。     

基于分形特征的云雾遥感图像分离方法*

雾是一种气象灾害,将雾从卫星云图上分离出来仍非易事。分数维给予图像纹理统计意义的描述,有效地体现了纹理的复杂度和粗糙度,揭示了纹理内在的自相似性,为云雾图像纹理分析提供了新的思路。简要阐述了纹理图像的差值盒维数计算方法,计算并分析了云雾纹理图像的差值盒维数特征。针对差值盒维数在表现云雾纹理特征和云雾分离方面存在的问题,提出基于样本图像灰度均值的加权盒维数算法,以改变出现灰值差异较大的不同云类具有相同盒维数的情况,并与云雾的光谱特征结合,实现雾与云的识别与分离,在实际应用     

基于三维盒维数算法的城市分维计算

通过将传统的二维盒维数算法扩展到三维空间,提出了一个基于三维空间的盒维数计算方法。分别利用三维盒维数算法和二维盒维数算法计算城市的分维,通过对计算结果进行比较分析,观察到城市空间结构在第三维同样具有分形特征,证明传统城市分维计算中采用基于二维空间的分维算法或者简单地利用二维分维加1的方法表示三维分维都是不准确的,并进而给出正确的城市分维计算方法。     

基于小波变换和分形维数的车牌汉字识别

提出一种基于小波变换与分形维数的车牌汉字识别方法.对字符图像进行预处理和小波变换,应用改进的微分盒维法计算图像分形盒维值,并构造特征向量,利用支持向量机分类器对字符进行分类与识别.实验结果表明,该方法对模糊字符的识别具有鲁棒性,可提高汉字识别率.     

基于分形维数的雷达信号脉内调制方式识别

分形维数是分形理论中的重要参数,其中盒维数可以描述分形信号的几何尺度信息,信息维数描述分形信号在平面空间上的分布信息。因此利用分形理论从信号中提取盒维数和信息维数作为分类特征,这种特征可包含信号幅度、频率和相位的变化规律,并集中了不同调制信号之间的差异信息。通过计算若干种常见脉内调制雷达辐射源信号的盒维数与信息维数,表明了分形维数在分类意义上是有效的特征。并由分类识别的仿真实验验证了基于分形维数进行的分类具有较高的识别率。     

基于PCA-SSA-BP神经网络的烟叶分级方法

基于机器视觉检测的烟叶分级方法,存在识别效率低、识别稳定性差等问题。针对这些问题,该文基于主成分分析（PCA）、麻雀搜索算法（SSA）、BP神经网络提出一种烟叶分级方法,首先对采集到的烟叶图像进行分析提取17个图像特征进行PCA降维处理,以消除冗余信息,结合烟叶分级标准并平衡烟叶分级的识别正确率和分级效率确定最佳降维的维数为七维;然后,将七维特征向量作为BP神经网络的输入,利用SSA对BP网络模型的权值和阈值进行优化,完成网络模型的训练。实验结果表明,PCA-SSA-BP模型的平均识别率达到96%以上,总运行时间为97.99 s,因此该方法能有效提高小样本烟叶分级的识别效率和识别稳定性。     

基于Gabor小波与分形维的人脸情感特征提取

提出一种基于Gabor小波变换与分形维的人脸情感特征提取算法,对包含情感信息的静态灰度图像进行预处理,对表情子区域实行Gabor小波变换,提取情感特征矢量,对人脸兴趣区图像求盒维数和差分分形维数,将经过Gabor小波变换所得的特征矢量和分形维数作为所提取的特征。分析比较了不同测试者7种基本情感的识别效果,实验表明该方法能有效提取与情感变化有关的特征。     

基于布朗模型和小波变换的图像分形维数计算

图像分形维数是反映图像纹理特征的重要因素,也是图像分割的主要依据;通常,图像的分形维数多数采用盒维数计算方法来得到,但是避免不了计算时阈值选择带来不精确的问题,本文结合小波变换和布朗模型,提出了一种新的计算方法,并且和盒维数方法计算结果进行比较,结果表明,通过本文的计算方法得到的图像分形维数较准确。     

A Fractal Method for Digital Elevation Model Construction and its Application to a Mountain Region

This paper presents and discusses the results of a methodology for the construction of Digital Elevation Models (DEM), based on mathematical principles of fractal geometry. Classical approaches to DEM are founded on numerical methods of interpolation or approximation of data extracted from contour lines of standard topographic maps. Fractal geometry opens a new and innovative approach in which the irregularity, erraticity and self similarity of fractal structures mimics at best the typical behaviour of high mountain profiles. The adoption of fractal methods requires the identification of the fractal dimension D for a given data set. The determination of the fractal dimension D can be achieved applying some special mathematical method on finite set of samples. Different methods. for D estimation are presented. Once the fractal dimension is given, it is possible to reconstruct a surface relief using different fractal methods which interpolate the original set of elevation data. Our research has also investigated two different interpolation methods, fBm and IFS, both confirming that the fractal approach is very well suited for high relief terrain modeling. Photo realistic rendering of the computed DEM allows to appreciate the quality of the reconstruction through visual inspection.     

基于分形维数的地表模型多分辨率动态绘制

以基于分形维数的树状结构组织三维地表网格模型,实现了与视点相关的连续多分辨率地表模型简化及实时绘制.算法分为预处理和实时绘制两个阶段.在预处理阶段,通过分形维数评价地表的复杂度,建立自适应的树状结构,计算出所有顶点的误差值.在绘制阶段,则根据视距、视角等因素动态地确定需保留的顶点集,并采用受限四叉树方法实时三角化得到所需分辨率下的三角网格近似模型.该算法具有两个优点:一个是地表模型的分层区域划分考虑了地形本身的复杂度;另一个是建立了视点相关各参数与所采用的分辨率表示的直接关系.实验表明,此算法简单、有效,支持对地表模型的交互式实时动态绘制.     

一种带钢表面缺陷识别与分类的研究——基于混合加权特征和RBF网络的方法

本论文中,采用灰度直方图特征、灰度共生矩阵特征和小波变换特征的提取方法,三种特征方法的结合能够很好的实现分类的目的。在提取特征向量的基础上,本研究基于MATLAB6.5环境下的神经网络工具箱,采用了兼顾识别速度与分类准确性的RBF神经网络分类器对带钢表面缺陷进行识别与分类,此算法可以作为高速生产线的带钢表面缺陷的实时检测优选方案。     

基于软件寻位的自由曲面零件超声检测方法

针对复杂曲面工件内部缺陷的超声检测,本文提出了一种基于软件寻位的超声检测轨迹生成算法,只需获取检测工件的少量信息,就能自动求解工件状态,从而实时生成检测轨迹.使用该方法无需使用精密夹具和人工找正,因而可减少产品成本,提高检测效率.     

Video Grading of Oranges in Real-Time

We describe a novel system for grading oranges into three quality bands, according to their surface characteristics. The system is designed to process fruit with a wide range of size (55–100 mm), shape (spherical to highly eccentric), surface coloration and defect markings. This application requires both high throughput (5–10 oranges per second) and complex pattern recognition. The grading is achieved by simultaneously imaging each item of fruit from six orthogonal directions as it is propelled through an inspection chamber. In order to achieve the required throughput, the system contains state-of-the-art processing hardware, a novel mechanical design, and three separate algorithmic components. One of the key improvements in this system is a method for recognising the point of stem attachment (the calyx) so that it can be distinguished from defects. A neural network classifier on rotation invariant transformations (Zernike moments) is used to recognise the radial colour variation that is shown to be a reliable signature of the stem region. The succession of oranges processed by the machine constitute a pipeline, so time saved in the processing of defect free oranges is used to provide additional time for other oranges. Initial results are presented from a performance analysis of this system.     

Inspection equipment study for subway tunnel defects by grey-scale image processing

In recent years, much attention has been paid to Machine Vision-Based (MVB) technology for tunnel main defect (leakage and crack) inspection as an innovative technology. Based on the principle of MVB technology, various researchers have developed tunnel inspection equipment, but most of them need either a trailer or an external power supply, which cannot meet the demand of subway tunnel inspection in China. The limited inspection time, high demand for precision, rigid requirements of operational management and high cost of the equipment restrict the application of this method in China. MTI-100 (Moving Tunnel Inspection) was developed under these circumstances. To capture stable, high-quality images of the lining surface as the raw data of inspection, an image capture system is well designed based on CCD (Charge-coupled Device) camera scanning. Additionally, equipment optimization design of the mechanism and electricity requirements for the inspection accuracy of subway tunnel inspection is investigated. The maximal size and weight of equipment elements determined the convenience of inspection, which is primarily conditioned by these designs. The effects of lighting and vibration have been considered. A method to calculate the image shift caused by vibration is proposed. The software network is another core component of the equipment, which connects the image acquisition, image storage and defect recognition. The famous Otsu method is used for leakage recognition. A new algorithm based on the features of the local image grid is developed to recognize cracks. A comparative study shows its high accuracy for crack recognition. Finally, a simulative tunnel test and field inspection are undertaken to verify the performance of the non-destructive subway tunnel inspection equipment. Through these tests, the accuracy, stability, repeatability, labor intensity and efficiency of the equipment have been verified. A real project test certified that the developed MTI-100 is quite suitable for practical tunnel inspection.     

Partial discharge pattern recognition of power cable joints using extension method with fractal feature enhancement

This paper proposes a new partial discharge (PD) pattern recognition using the extension method with fractal feature enhancement. First, four common defect types of XLPE power cable joints are established, and a commercial PD detector is used to measure the PD signal by inductive sensor (L-sensor). Next, the feature parameters of fractal theory (fractal dimension and lacunarity) are extracted from the 3D PD patterns. Finally, the matter-element models of the PD defect types are built. The PD defect types can be directly identified by the degree of correlation between the tested pattern and the matter-element based on the extension method. The extension method needs representative features to define the interval of the matter-element. In order to enhance the extension performance, we add fractal features that are extracted from the PD 3D patterns. To demonstrate the effectiveness of the extension method with fractal feature enhancement, the identification ability is investigated on 120 sets of field-tested PD patterns of XLPE power cable joints. Compared with the back-propagation neural network (BPNN) method, the results show that the extension method with fractal feature enhancement not only has high recognition accuracy and good tolerance when random noise is added, but that it also provides fast recognition speed.     

分形特性角度定位研究

提出一种将分形理论用于PCB图像定位的新方法，以PCB上的矩形定位标志图像为研究对象，分析其边缘特征，结果表明图像边缘具有典型的分形特性；用去除边缘异常噪声加补偿的方法对图像边缘进行处理，消除噪声对图像边缘对应的分形曲线特性的影响，得到较精确的边缘分形曲线形维数；推导出图像转动角度与图像边缘分开曲线分形维数的理论关系式，根据边缘曲线的分形维数精确地得到了目标图像的转动角度，解决了PCR板的精确，快速角度定位问题。     

Neural network based detection of local textile defects

A new approach for the segmentation of local textile defects using feed-forward neural network is presented. Every fabric defect alters the gray-level arrangement of neighboring pixels, and this change is used to segment the defects. The feature vector for every pixel is extracted from the gray-level arrangement of its neighboring pixels. Principal component analysis using singular value decomposition is used to reduce the dimension of feature vectors. Experimental results using this approach illustrate a high degree of robustness for the detection of a variety of fabric defects. The acceptance of a visual inspection system depends on economical aspects as well. Therefore, a new low-cost solution for the fast web inspection using linear neural network is also presented. The experimental results obtained from the real fabric defects, for the two approaches proposed in this paper, have confirmed their usefulness.     

New method for feature extraction based on fractal behavior

In this paper, a novel approach to feature extraction based on fractal theory is presented as a powerful technique in pattern recognition. This paper presents a new fractal feature that can be applied to extract the feature of two-dimensional objects. It is constructed by a hybrid feature extraction combining wavelet analysis, central projection transformation and fractal theory. New fractal feature and fractal signatures are reported. A multiresolution family of the wavelets is also used to compute information conserving micro-features. We employed a central projection method to reduce the dimensionality of the original input pattern. A wavelet transformation technique to transform the derived pattern into a set of sub-patterns. Its fractal dimension can readily be computed, and to use the fractal dimension as the feature vectors. Moreover, a modified fractal signature is also used to distinguish the distinct handwritten signatures. We expect that the proposed fractal method can also be used for improving the extraction and classification of features in pattern recognition.