全自动划片机自动识别对准技术的研究

主要阐述了数字图像处理技术在全自动划片机中的应用,简单的分析了自动识别对准技术的原理及实现手段,为其他电子专用设备中自动识别对准的应用提供一些思路与技巧。     

基于核机器学习的腔体滤波器辅助调试

 研究了一种基于核机器学习的腔体滤波器辅助调试方法.该方法根据工程中的调试经验数据,首先使用核机器学习算法建立了螺栓调整量和滤波器电性能之间的影响关系模型.然后应用此模型,使用优化技术构建了滤波器的辅助调整方法.实际滤波器的实验结果表明了该方法的有效性.该方法比较适用于工程中批量生产的腔体滤波器的辅助调试.     

用于手背静脉注射的图像自动识别与标注

针对医疗领域的手背静脉识别研究甚少的问题,提出一种用于静脉注射的手背静脉自动识别的方法。分析了静脉穿刺的特点以及适合注射的静脉条件,在近红外光下获得全局手背静脉图像的基础上,首先对图像进行预处理获得完整的细化图像,再利用改进的区域生长法分别对每条静脉进行提取和分析,最后用模板匹配法识别选择最适合穿刺的静脉部分。实验结果表明：该方法能够实现对适合注射的静脉部分进行自动识别与标注。     

基于机器学习的网络感知评估方法

提升网络感知和客户满意度一直是网络优化的工作主线,而KPI指标无法反映网络真实感知情况,传统通过调研了解客户满意度的方式存在很大局限性。本文深入研究了KPI指标和网络真实感知的映射关系,通过大数据挖掘和机器学习建模实现了感知权重因子的量化,以此为基础完成了一种基于机器学习的网络感知评估方法,为客户满意度提升工作提供了全新的分析思路和支撑手段。     

基于机器学习的激光遥感图像多目标检测方法

目前方法检测多目标激光遥感图像时,其未处理遥感图像的辐射量失真情况,导致检测方法的平均识别率、平均精度均值低和检测效率低。提出基于机器学习的激光遥感图像多目标检测方法,该方法处理激光遥感图像辐射量失真情况,通过校正辐射和辐射匹配获取到更清晰和真实的遥感图像,结合改进的原始Mask R-CNN网络,引入分级跳连法和K均值聚类算法,解决Mask R-CNN网络不适用于激光遥感图像检测的问题,采用改进的机器学习,实现激光遥感图像多目标检测。实验结果表明,所提方法检测出激光遥感图像中的全部目标,平均识别率为95.6%,目标检测时间为0.02 s,平均精度均值达到77.5%,因此,该方法有效提高了平均识别率、平均精度均值和检测效率。     

脉冲压缩雷达抗电子攻击性能研究

脉冲压缩技术是雷达与干扰进行功率对抗的一种有效的技术手段,本文详细的分析了脉冲压缩的原理和信号特点,对攻击脉冲压缩雷达的方法进行了研究。     

调频连续波的一种匹配滤波处理方法

脉冲压缩雷达中广泛使用调频连续波形,使用匹配滤波的处理方法。但对大时宽带宽积的任意调频连续波其匹配滤波器的设计是比较困难的。该文提出一种比较易于实现的匹配滤波处理方法,具有较大的实用价值。     

复杂背景下的汽车牌照自动识别系统

分析了目前存在的车牌自动识别系统及其相应的问题,提出了一种在多分辨率下串行运算的金字塔结构的车牌自动识别系统.该系统由两个重要步骤组成:1)基于灰度方差的车牌定位算法,2)结合连通域分析和投影分析的字符切分算法.通过大量的实验证明该方法准确率高,鲁棒性好,速度快.     

基于MATLAB线性平滑滤波器的设计

本文采用MATLAB透行线性平滑滤波器设计.结果表明,线性低通平滑滤波器可以有效的消除嗓声,当所用的平滑模板的尺寸增大时,消除嗓声的效果增强,同时所得的图像变得模糊,细节的锐化提度逐步减弱.     

雷达中的相位编码信号与处理

介绍雷达中常用巴克码、弗兰克码、P码等相位编码信号.阐述它们的特点及在雷达中的应用,并进一步介绍对于这类信号的典型处理方法.     

Automatic CRP mapping using nonparametric machine learning approaches

This paper studies an uneven two-class unsupervised classification problem of satellite imagery, i.e., the mapping of U.S. Department of Agriculture's (USDA) Conservation Reserve Program (CRP) tracts. CRP is a nationwide program that encourages farmers to plant long-term, resource conserving covers to improve soil, water, and wildlife resources. With recent payments of nearly US $1.6 billion for new enrollments (2002 signup), it is imperative to obtain accurate digital CRP maps for management and evaluation purposes. CRP mapping is a complex classification problem where both CRP and non-CRP areas are composed of various cover types. Two nonparametric machine learning approaches, i.e., decision tree classifier (DTC) and support vector machine (SVMs) are implemented in this work. Specifically, considering the importance of CRP classification sensitivity, a new DTC pruning method is proposed to increase recall. We also study two SVM relaxation approaches to increase recall. Moreover, a localized and parallel framework is suggested in order to efficiently deal with the large-scale CRP mapping need. Simulation results validate the applicability of the suggested framework and proposed techniques.     

Robotic grasping recognition using multi-modal deep extreme learning machine

Recognizing which part of an object is graspable or not is important for intelligent robot to perform some complicated tasks. In order to obtain good grasping performance, learning rich representations efficiently from multi-modal RGB-D images is crucial. To address this problem, in this paper, we propose an effective multi-modal deep extreme learning machine structure. In this structure, unsupervised hierarchical extreme learning machine (ELM) is conducted for feature extraction for RGB and depth modalities separately. Then, the shared layer is developed by combining both RGB and depth features. Finally, the ELM is used as supervised feature classifier for final decision. Experimental validation on Cornell grasping dataset illustrates that the proposed multiple modality fusion method achieves better grasp recognition performance.     

具有学习功能的自动人脸识别

人脸识别是模式识别领域中一个相当困难而又有理论意义和实际价值的研究课题。传统的基于K－L变换的自动人脸识别方法,不用过多地考虑人脸的局部特征,利用特征脸方法进行识别,取得了一定的。但是,人脸作为一个特殊的场景,脸像会受年龄、心情、拍摄角度、光照条件、发饰等因素影响,所成图像存在差异。传统的基于K－L变换的自动人脸识别方法不能很好地克服这些畸变的影响。文中就主成分分析方法引入人脸识别,模拟人脸脸像的各种变化,事先对脸像做相应的变化,产生一系列变形脸。然后对变形脸进行主成分分析,提取它们的主成分。最后应用遗传算法选择最优特征向量构造子空间,提出一种能抗御一定脸像变化的人脸识别方法,并运用该方法进行了实验。实验结果证明了该方法的可行性和良好的抗畸变能力。     

An efficient method for handling ectopic beats using the heart timing signal

The problem of analyzing heart rate variability in the presence of ectopic beats is revisited. Based on the integral pulse frequency modulation model and the closely related heart timing signal, a new technique is introduced which corrects for the occasional presence of ectopic beats. The correction technique, which involves the occurrence times of a certain number of beats preceding the ectopic beat, is computationally very efficient. From actual heart rate data, the results show that the new technique is associated with a much lower computational complexity (flops reduced by a factor of about 3000) than the original heart timing technique, while producing similar performance. It is also shown that the power spectrum and related clinical indices obtained by the new technique are more accurately estimated than by other methods.     

基于支持向量机的语种识别

支持向量机是统计学习理论的一个重要分支,也是解决模式识别问题的一个有力工具.现简要介绍支持向量机理论,构建基于径向基函数的支持向量机,对分别来自两个不同的电话语音数据库中的汉语普通话、英语、日语、白族语和纳西语等5种语言进行识别研究.实验结果表明,支持向量机, 对不同数据库语言的语种识别依然能达到比较高的识别率.     

小波核极限学习机在人脸识别中的应用

文章提出了一种基于小波核极限学习机(Wavelet Kernel Extreme Learning Machine,WK-ELM)的人脸识别算法。首先,使用2D盖博小波变换对人脸图片进行初步的人脸特征提取。为了从所有提取的特征中选择出与人脸识别相关的、必要的特征,使用主成分分析法(Principal Component Analysis,PCA)对经过初步处理后的图像再进行进一步处理,有效地降低了特征维数。然后使用小波核极限学习机对提取到的图像进行分类。实验证明,小波核极限学习机不仅识别性能高,而且训练速度也优于其他算法。     

Automatic and objective gradation of 114 183 terrorist attacks using a machine learning approach

Catastrophic events cause casualties, damage property, and lead to huge social impacts. To build common standards and facilitate international communications regarding disasters, the relevant authorities in social management rank them in subjectively imposed terms such as direct economic losses and loss of life. Terrorist attacks involving uncertain human factors, which are roughly graded based on the rule of property damage, are even more difficult to interpret and assess. In this paper, we collected 114 183 open-source records of terrorist attacks and used a machine learning method to grade them synthetically in an automatic and objective way. No subjective claims or personal preferences were involved in the grading, and each derived common factor contains the comprehensive and rich information of many variables. Our work presents a new automatic ranking approach and is suitable for a broad range of gradation problems. Furthermore, we can use this model to grade all such attacks globally and visualize them to provide new insights.     

Prerequisites between learning objects: Automatic extraction based on a machine learning approach

One standing problem in the area of web-based e-learning is how to support instructional designers to effectively and efficiently retrieve learning materials, appropriate for their educational purposes. Learning materials can be retrieved from structured repositories, such as repositories of Learning Objects and Massive Open Online Courses; they could also come from unstructured sources, such as web hypertext pages. Platforms for distance education often implement algorithms for recommending specific educational resources and personalized learning paths to students. But choosing and sequencing the adequate learning materials to build adaptive courses may reveal to be quite a challenging task.In particular, establishing the prerequisite relationships among learning objects, in terms of prior requirements needed to understand and complete before making use of the subsequent contents, is a crucial step for faculty, instructional designers or automated systems whose goal is to adapt existing learning objects to delivery in new distance courses. Nevertheless, this information is often missing. In this paper, an innovative machine learning-based approach for the identification of prerequisites between text-based resources is proposed. A feature selection methodology allows us to consider the attributes that are most relevant to the predictive modeling problem. These features are extracted from both the input material and weak-taxonomies available on the web. Input data undergoes a Natural language process that makes finding patterns of interest more easy for the applied automated analysis. Finally, the prerequisite identification is cast to a binary statistical classification task. The accuracy of the approach is validated by means of experimental evaluations on real online coursers covering different subjects.     

Analysis of heart rate variability in the presence of ectopic beats using the heart timing signal

The time-domain signals representing the heart rate variability (HRV) in the presence of an ectopic beat exhibit a sharp transient at the position of the ectopic beat, which corrupts the signal, particularly the power spectral density (PSD) of the HRV. Consequently, there is a need for correction of this type of beat prior to any HRV analysis. This paper deals with the PSD estimation of the HRV by means of the heart timing (HT) signal when ectopic beats are present. These beat occurrence times are modeled from a generalized, continuous time integral pulse frequency modulation model and, from this point of view, a specific method for minimizing the effect of the presence of ectopic beats is presented to work together with the HT signal. By using both, a white noise driven autoregressive model of the HRV signal with artificially introduced ectopic beats and actual heart rate series including ectopic beats, the more usual methods of HRV spectral estimation are compared. Results of the PSD estimation error function of the number of ectopic beats are presented. These results demonstrate that the proposed method has one order of magnitude lower error than usual ectopic beats removal strategies in preserving PSD, thus, this strategy better recovers the original clinical indexes of interest.