错位指纹图像自动检测的深度学习方法

针对实践中存在的指纹图像错位问题,提出了基于深度学习的错位指纹图像自动识别算法.通过将错位指纹自动检测问题转化为一个四分类问题,搭建了一个深度卷积神经网络并用已知的标签进行有监督学习,将学习得到的模型用于预测给定指纹图像的类别.通过对模型性能的准确率、空图识别率和错位判正率3项指标的评价看出,本识别算法取得了良好的效果.     

利用B样条曲线设计指纹图像滤波模板和快速指纹特征匹配算法

在指纹图像识别过程中,指纹图像的滤波去噪音二值化和指纹特征的匹配是最关键的两个部分.本文针对这两部分的算法设计进行分析和改进,介绍利用B样条造型来设计指纹滤波模板的方法和基于基准点附近的拓扑结构进行快速特征匹配的算法.并且利用改进后的指纹特征匹配算法,在样本指纹库中进行实验,结果表明该算法大大提高了指纹特征匹配的效率和准确性.     

基于超声同步观测辅助技术的CT成像系统设计及图像融合

CT检测非刚性物体目标时,会因多角度扫描的异步性而使得重建图像分辨率降低.本文利用超声实时同步观测技术,设计了一种基于超声同步观测辅助技术的CT成像系统,使得通过该成像系统能获取同一个标定系统下的CT重建图像和超声图像.同时基于此系统提出了一个区域特征加权的图像融合方案,从而达到丰富重建图像信息量和提高信息准确率的目的.     

基于深度残差网络的水下海参自动识别研究

水下海参自动识别是水下机器人捕捞海参的关键技术,为解决水下海参识别,基于深度残差网络构建了水下海参识别自动识别算法.研究图像数据采集自山东省海阳市千岛湖海参养殖场,以海参图像为正样本,非海参图像为负样本构建数据集,分别以图像分辨率32×32像素、经过对比度受限自适应直方图均衡化(Contrast-Limited Adaptive Histogram Equalization,CLAHE)算法增强的图像、分辨率为64×64像素图像、分辨率为128×128像素图像进行模型训练与识别.结果显示:1)经过CLAHE增强的图像在训练时效果较好,收敛速度快,准确性较高,但是经过训练的模型泛化性能差;2)图像分辨率64×64像素时,在收敛速度、准确性方面更优,识别准确度、精确度、召回率分别为0.9037、0.9143、0.8889,AUC为0.95;3)图像分辨率为128×128像素时,识别效果最差,图像分辨率的提高没有提高模型的识别准确率.     

基于3G网络的河道冰凌图像遥测系统设计与冰凌密度分析

设计了一种适用于野外无供电条件下的河道冰凌图像远程遥测系统,介绍了系统的整体结构、运行模式和关键电路的构成以及图像处理模块的选型.设计系统可以通过3G无线网络将上位微机图像监测控制命令传送到河道现场,可以实时控制现场控制摄像头的旋转角度、方向及焦距,实现对被监测河道区域冰凌图像的多视角实时在线监测;系统上位微机管理软件对接收到的冰凌图像信号经过动态阈值分割算法处理,可获得观测区域河道水面冰凌密度数值.     

高中课程标准下不等式性质教学内容安排的思考

当前,高中课程改革正在全国各省市逐步推开.新课程标准关于不等式的内容与以前的教学大纲相比有了明显的改变.原有的主要内容分别归人了必修模块5和选修模块4-5,归并、增添了一些新的内容.因此,不等式教学面临着明显的变化.其中不等式的性质在教学内容上课程标准未给出具体要求,使得不同版本的教材之间形成了一定的差异,也对教学的实施带来了不确定因素和几个相关的问题.     

图像反问题中的数学与深度学习方法

我们生活在数字的时代,数据已经成为了我们生活中不可或缺的一部分,而图像无疑是最重要的数据类型之一.图像反问题,包括图像降噪,去模糊,修复,生物医学成像等,是图像科学中的重要领域.计算机技术的飞速发展使得我们可以用精细的数学和机器学习工具来为图像反问题设计有效的解决方案.本文主要回顾图像反问题中的三大类方法,即以小波（框架）为代表的计算调和分析法、偏微分方程（PDE）方法和深度学习方法.我们将回顾这些方法的建模思想和一些具体数学形式,探讨它们之间的联系与区别,优点与缺点,探讨将这些方法有机融合的可行性与优势.     

基于改进遗传算法的自动阈值图像分割方法

图像分割技术在图像分析和图像识别上具有重要意义.传统自适应遗传算法有可能使问题求解陷入局部最优解,而求得错误的图像分割阈值.为了得到最优的图像分割阈值,对交叉率和变异率公式进行了重构,使得交叉率和变异率在任何情况下都不为零.同时,以最大二维熵函数作为适应度函数,采用选择、交叉变异等遗传操作作搜索最优分割阈值.仿真实验表明,该方法可以有效地提高图像分割精度和计算速度.     

一种改进的混沌序列产生方法

根据灰度图像的二维直方图的特点,在已有的二维Arnold混沌系统的基础上,结合Bernstein形式的Bézier曲线的生成算法,给出了一种基于生成Bézier曲线的de Casteljau算法构造伪随机序列的方法,实验结果表明生成的二维序列不仅具有伪随机性,而且还具有在近似圆盘中随机分布的性质,这使得该伪随机序列更适合对灰度图像的二维灰度直方图进行基于混沌优化的图像分割.在此基础上,给出了一种基于混沌优化的二维最大熵的灰度图像分割算法,该算法对于含噪图像取得了良好的分割效果.     

基于改进Gentle-Adaboost算法的疲劳驾驶中人脸检测研究

疲劳驾驶检测中人脸检测的算法有很多,其中相对成熟的是Gentle-Adaboost,此算法虽然识别率较高,但是识别时间较长,研究在识别过程中使用均值哈希算法进行算法优化,并引入缓存的概念来缩短识别时间.首先,根据Gentle-Adaboost算法和Haar特征来提取基本特征点,然后通过对比前后两帧图像的哈希指纹来减少人脸检测次数,最后通过缓存数据库来存储相似图片的哈希指纹,运行一段时间后仅需通过对比哈希指纹就能精准的找到人脸区域.通过实验,可以证明改进后的平均识别时间可减少原时间的80%.     

Solving a k-Node Minimum Label Spanning Arborescence Problem to Compress Fingerprint Templates

We present a novel approach for compressing relatively small unordered data sets by means of combinatorial optimization. The application background comes from the field of biometrics, where the embedding of fingerprint template data into images by means of watermarking techniques requires extraordinary compression techniques. The approach is based on the construction of a directed tree, covering a sufficient subset of the data points. The arcs are stored via referencing a dictionary, which contains “typical” arcs w.r.t. the particular tree solution. By solving a tree-based combinatorial optimization problem we are able to find a compact representation of the input data. As optimization method we use on the one hand an exact branch-and-cut approach, and on the other hand heuristics including a greedy randomized adaptive search procedure (GRASP) and a memetic algorithm. Experimental results show that our method is able to achieve higher compression rates for fingerprint (minutiae) data than several standard compression algorithms.     

Quality-augmented fusion of level-2 and level-3 fingerprint information using DSm theory

Existing algorithms that fuse level-2 and level-3 fingerprint match scores perform well when the number of features are adequate and the quality of images are acceptable. In practice, fingerprints collected under unconstrained environment neither guarantee the requisite image quality nor the minimum number of features required. This paper presents a novel fusion algorithm that combines fingerprint match scores to provide high accuracy under non-ideal conditions. The match scores obtained from level-2 and level-3 classifiers are first augmented with a quality score that is quantitatively determined by applying redundant discrete wavelet transform to a fingerprint image. We next apply the generalized belief functions of Dezert–Smarandache theory to effectively fuse the quality-augmented match scores obtained from level-2 and level-3 classifiers. Unlike statistical and learning based fusion techniques, the proposed plausible and paradoxical reasoning approach effectively mitigates conflicting decisions obtained from classifiers especially when the evidences are imprecise due to poor image quality or limited fingerprint features. The proposed quality-augmented fusion algorithm is validated using a comprehensive database which comprises of rolled and partial fingerprint images of varying quality with arbitrary number of features. The performance is compared with existing fusion approaches for different challenging realistic scenarios.     

Computational investigations of maximum flow algorithms

The maximum flow algorithm is distinguished by the long line of successive contributions researchers have made in obtaining algorithms with incrementally better worst-case complexity. Some, but not all, of these theoretical improvements have produced improvements in practice. The purpose of this paper is to test some of the major algorithmic ideas developed in the recent years and to assess their utility on the empirical front. However, our study differs from previous studies in several ways. Whereas previous studies focus primarily on CPU time analysis, our analysis goes further and provides detailed insight into algorithmic behavior. It not only observes how algorithms behave but also tries to explain why algorithms behave that way. We have limited our study to the best previous maximum flow algorithms and some of the recent algorithms that are likely to be efficient in practice. Our study encompasses ten maximum flow algorithms and five classes of networks. The augmenting path algorithms tested by us include Dinic's algorithm, the shortest augmenting path algorithm, and the capacity-scaling algorithm. The preflow-push algorithms tested by us include Karzanov's algorithm, three implementations of Goldberg-Tarjan's algorithm, and three versions of Ahuja-Orlin-Tarjan's excess-scaling algorithms. Among many findings, our study concludes that the preflow-push algorithms are substantially faster than other classes of algorithms, and the highest-label preflow-push algorithm is the fastest maximum flow algorithm for which the growth rate in the computational time is O(n^1.5) on four out of five of our problem classes. Further, in contrast to the results of the worst-case analysis of maximum flow algorithms, our study finds that the time to perform relabel operations (or constructing the layered networks) takes at least as much computation time as that taken by augmentations and/or pushes.     

Neighborhood-following algorithms for linear programming

In this paper, we present neighborhood-following algorithms for linear programming. When the neighborhood is a wide neighborhood, our algorithms are wide neighborhood primal-dual interior point algorithms. If the neighborhood degenerates into the central path, our algorithms also degenerate into path-following algorithms. We prove that our algorithms maintain the O(n~(1/2)L)-iteration complexity still, while the classical wide neighborhood primal-dual interior point algorithms have only the O(nL)-iteration complexity. We also proved that the algorithms are quadratic convergence if the optimal vertex is nondegenerate. Finally, we show some computational results of our algorithms.     

On the Solution of Linear Programs by Jacobian Smoothing Methods

We introduce a class of algorithms for the solution of linear programs. This class is motivated by some recent methods suggested for the solution of complementarity problems. It reformulates the optimality conditions of a linear program as a nonlinear system of equations and applies a Newton-type method to this system of equations. We investigate the global and local convergence properties and present some numerical results. The algorithms introduced here are somewhat related to the class of primal–dual interior-point methods. Although, at this stage of our research, the theoretical results and the numerical performance of our method are not as good as for interior-point methods, our approach seems to have some advantages which will also be discussed in detail.     

The convergence of V-cycle multigrid algorithms for axisymmetric Laplace and Maxwell equations

We investigate some simple finite element discretizations for the axisymmetric Laplace equation and the azimuthal component of the axisymmetric Maxwell equations as well as multigrid algorithms for these discretizations. Our analysis is targeted at simple model problems and our main result is that the standard V-cycle with point smoothing converges at a rate independent of the number of unknowns. This is contrary to suggestions in the existing literature that line relaxations and semicoarsening are needed in multigrid algorithms to overcome difficulties caused by the singularities in the axisymmetric Maxwell problems. Our multigrid analysis proceeds by applying the well known regularity based multigrid theory. In order to apply this theory, we prove regularity results for the axisymmetric Laplace and Maxwell equations in certain weighted Sobolev spaces. These, together with some new finite element error estimates in certain weighted Sobolev norms, are the main ingredients of our analysis.

     

On Some Techniques for Streaming Data: A Case Study of Internet Packet Headers

We consider the implications of streaming data for data analysis and data mining. Streaming data are becoming widely available from a variety of sources. In our case we consider the implications arising from Internet traffic data. By implication, streaming data are unlikely to be time homogeneous so that standard statistical and data mining procedures do not necessarily apply. Because it is essentially impossible to store streaming data, we consider recursive algorithms, algorithms which are adaptive and discount the past and also algorithms that create finite pseudo-samples. We also suggest some evolutionary graphics procedures that are suitable for streaming data. We begin our discussion with a discussion of Internet traffic in order to give the reader some sense of the time and data scale and visual resolution needed for such problems.     

Approximation algorithms for nonnegative polynomial optimization problems over unit spheres

We consider approximation algorithms for nonnegative polynomial optimization problems over unit spheres. These optimization problems have wide applications e.g., in signal and image processing, high order statistics, and computer vision. Since these problems are NP-hard, we are interested in studying on approximation algorithms. In particular, we propose some polynomial-time approximation algorithms with new approximation bounds. In addition, based on these approximation algorithms, some efficient algorithms are presented and numerical results are reported to show the efficiency of our proposed algorithms.     

New Algorithms of Distance-Increasing Mappings from Binary Vectors to Permutations by Swaps

Mappings from the set of binary vectors of a fixed length to the set of permutations of the same length that strictly increase Hamming distances are useful for the construction of permutation codes (permutation arrays). In this paper, we propose new simpler algorithms of distance-increasing mappings. These algorithms do not need any table lookup operations, and they are built up with fewer swap perations. In the comparison of our new algorithms with other DIMs, we also give some numerical results to illustrate that the distance expansion distributions of our new mappings are not bad.     

Ridge directional singular points for fingerprint recognition and matching

In this paper, a new approach to extract singular points in a fingerprint image is presented. It is usually difficult to locate the exact position of a core or a delta due to the noisy nature of fingerprint images. These points are the most widely used for fingerprint classification and matching. Image enhancement, thinning, cropping, and alignment are used for minutiae extraction. Based on the Poincaré curve obtained from the directional image, our algorithm extracts the singular points in a fingerprint with high accuracy. It examines ridge directions when singular points are missing. The algorithm has been tested for classification performance on the NIST‐4 fingerprint database and found to give better results than the neural networks algorithm. Copyright © 2005 John Wiley & Sons, Ltd.