用于EVS的改进小波变换图像融合算法

为增强小波变换图像融合算法的实时性,提高视觉增强系统(EVS)可见光图像与红外图像实时融合的效率,提出了一种基于矩阵QR分解和小波变换的图像融合算法.该算法对原始图像的像素矩阵进行QR分解,再利用正交矩阵的性质,根据小波变换图像融合算法对QR分解得到的上三角矩阵进行分解融合,利用QR分解得到的正交矩阵逆变换得到融合图像.实验结果表明,该算法能获得较好的实时性,同时保证较好的融合效果.     

归一化积相关图像匹配算法中的图像分块并行处理方法

对具有良好匹配特性的归一化积相关算法中的相关求和 - ∑∑XY的并行算法进行了较为深入的探讨 ,设计了基于 K元 2立方体网络计算机的图像分块计算 ∑∑XY的并行算法 .对于 N× N的参考图像、M× M的实时图像和K× K的处理元阵列 ( N=BK,K=M,B>1) ,通过将参考图像分成 B× B个图像块就可利用该算法实现归一化积相关图像分块匹配并行算法 .应用实验表明 ,本文设计的并行算法具有很好的并行效率     

一种基于深度学习的交通标志识别算法研究

针对当前在真实环境中交通标志呈多尺度分布,且图像背景复杂、天气光照多变等多种因素造成识别精度低、识别速度慢等情况.提出了一种基于深度学习神经网络的交通标志识别的设计与实现.首先从公开数据集TT100K中选取出现次数最多的45类交通标志进行识别,接着对图像进行mosaic等图像增强及图像处理.然后在深度学习神经网络中的Y...     

基于K均值聚类分割彩色图像算法的改进

基于人类视觉将图像分割成若干个有意义的区域是目标检测和模式识别的基础.应用K均值聚类算法对图像进行分析,分析了图像的空间、色彩以及纹理特征对聚类效果的影响,针对K均值算法的存在的过分割问题提出了一种修正方法,先基于空间、颜色和纹理特征分割图像,再基于色彩及纹理特征进行合并,解决了K均值聚类产生的过分割问题,并在区域合并时引入修正函数,抑制了图像中因场景明暗变化而产生的斑点.实验结果表明提出的聚类算法对图像分割效果有明显提高.     

基于压缩图像的K均值优化聚类相关反馈

在传统的基于内容的图像检索方法中,直接采集原始格式的图像检索比较多,由于数据量大,给存储或传输带来不便.基于小波变换和粒子群的K值聚类相关反馈的图像检索优点在于解决了数据量大、省略解压缩环节、特征向量包含在压缩域检索系数中;并克服了传统的K均值算法易陷入局部极小值的缺点.实验结果表明,图像压缩后的基于粒子群的K均值聚类提高了检索效率.     

心脏MRI图像快速分割方法

在分析心脏MR图像特点的基础上,提出了先对心脏MRI图像进行K均值聚类,把K均值聚类后的图像作为特征图像,在特征上用Song和Chan提出的快速分割方法进行粗分割,再用粗分割的曲线作为水平集的初始曲线,在心脏MRI图像上用Chan和Vese方法进行细分割的心脏MR图像分割方法.并对Song和Chan快速算法中扫描图像的区域进行了改进,提高了分割速度.分割实验证明,用该方法能够快速、准确地分割心脏MRI图像.     

图像缩放的研究与FPGA设计

针对图像缩放的不同要求,提出一种分步插值图像缩放器结构.对行列在空间上分开,在时间上同步,在开始的K(选择不同的算法K值不同)行行列串行处理,先进行行缩放,再用新生成的像素点进行列操作,K行处理完后,行列并行处理.这种结构使得图像缩放时并行度得到提升,由于行列分开处理使得对于不同要求的场合行列可以采用相同或不同的算法以达到特定的要求.实验表明,该分步式插值电路架构既节省了资源、降低了开销,又提高了图像质量.     

基于局部窗口K分布的快速舰船检测算法

针对局部窗口K分布检测算法运算速度慢、计算效率低的问题,提出了一种基于局部窗口K分布的快速舰船目标检测算法。该算法首先采用迭代分割算法对原始合成孔径雷达(SAR)图像进行预筛选处理,根据预筛选选出潜在目标,在原始SAR图像中剔除潜在目标像素;然后利用背景图像计算二阶和四阶积分图像,在每一个像素点处采用滑动窗口的方式,在积分图像中进行加减计算确定所在位置的二四阶矩并估计K分布的参数;其次,确定概率密度函数后,通过求解函数得到检测阈值,根据检测阈值确定感兴趣区域;最后,通过模糊差影的鉴别方法对目标中的虚警目标进行进一步剔除,进而完成检测。通过实测SAR图像检测实验,积分算法与局部窗口的K分布算法相比将运算所需时间降低了50%,基于模糊差影的鉴别算法将品质因素由44.4%提高到100%。所提算法既保证了算法的实时性,又提高了检测的精度,在进行SAR舰船自动检测方面具有一定的应用价值。     

基于微粒群的K均值聚类算法在图像分类中的应用

提出一种新的图象分类算法椈谖⒘Ｈ旱腒均值聚类图象分类算法.将此算法和K均值聚类算法以及微粒群图像分类算法分别应用于MRI人脑图象的分类,并进行了比较.实验结果表明:基于微粒群的K均值聚类图象分类算法具有较好的全局收敛性,不仅能有效克服K均值算法易陷入局部极小值的缺点,且全局收敛性能优于微粒群图像分类算法.     

K近邻优化估计的SAR图像建模与目标检测算法

在非均匀杂波环境下的合成孔径雷达(synthetic aperture radar,SAR)图像背景建模问题中,针对非参量建模算法Parzen窗估计严重依赖于窗宽设置及最优核函数选择的问题,提出一种基于K近邻优化的概率密度函数估计算法,解决因固定近邻数而导致估计不准确甚至不能估计的问题.该算法不需要图像的任何先验知识,且无需考虑窗宽的设置及最优核函数的选择问题.与Parzen窗估计、K分布和$G^0$分布的对比实验表明,所提出的K近邻优化估计算法可以实现对单峰、多峰甚至不规则图像数据的准确建模,优于K分布和$G^0$分布;同时,对图像首尾数据的处理优于Parzen窗估计.实验结果验证了所提出方法对SAR图像杂波建模的精确性、鲁棒性和简便性,以及全局恒虚警率目标检测的有效性.     

The association between students' use of an electronic voting system and their learning outcomes

Abstract This paper reports on the use of an electronic voting system (EVS) in a first-year computing science subject. Previous investigations suggest that students' use of an EVS would be positively associated with their learning outcomes. However, no research has established this relationship empirically. This study sought to establish whether there was an association between students' use of an EVS over one semester and their performance in the subject's assessment tasks. The results from two stages of analysis are broadly consistent in showing a positive association between EVS usage and learning outcomes for students who are, relative to their class, more correct in their EVS responses. Potential explanations for this finding are discussed as well as modifications and future directions of this program of research.     

‘Pretty Lights’ and Maths! Increasing student engagement and enhancing learning through the use of electronic voting systems

University classes in Mathematics are traditionally perceived to be uninspiring and devoid of active student–lecturer communication. Large undergraduate classes further compound the difficulty of engaging students and enabling viable student–lecturer feedback. At the Mathematics Education Centre, Loughborough University, some staff members have been using electronic voting systems (EVS) to enliven the classroom and enable large numbers of students to respond to questions in real time during class. In this paper, we present an evaluation case study, based on student perceptions, of the impact of EVS use on student learning and engagement. The results show that majority of students are hugely positive about the usefulness and overall advantageousness of EVS use in classes. Results also show that EVS use does increase the likelihood of students participating and engaging in class, as even students who do not view EVS as being particularly useful stated that they are more likely to participate in classes where EVS are used than otherwise. However, there seems to be no correlation between EVS use and improvement (or otherwise) in student grades.     

支持扩展虚拟同步的组通信算法形式描述与验证

组通信系统是为方便开发容错的分布式应用系统而提出的一种通信中间件．虚拟同步是组通信系统中的一个重要概念．其本质是限制向所有组成员递交组成员资格变化信息和应用消息的次序．为支持网络可划分的情况，引入了扩展虚拟同步模型．针对扩展虚拟同步模型的特点，提出了一种基于客户／服务器模式的组通信系统架构，并以I／O自动机的形式给出系统内部各模块的服务和算法．最后以继承建模的方式逐步给出该算法的自动机模型，并用形式化的方法验证其正确性．     

Examining the Effects of Conformal Terrain Features in Advanced Head‐Up Displays on Flight Performance and Pilot Situation Awareness

Synthetic vision systems (SVS) render terrain features for pilots through cockpit displays using a GPS database and three‐dimensional graphical models. Enhanced vision systems (EVS) present infrared imagery of terrain using a forward‐looking sensor in the nose of an aircraft. The ultimate goal of SVS and EVS technologies is to support pilots in achieving safety under low‐visibility and night conditions comparable to clear, day conditions. This study assessed pilot performance and situation awareness (SA) effects of SVS and EVS imagery in an advanced head‐up display (HUD) during a simulated landing approach under instrument meteorological conditions. Videos of the landing with various HUD configurations were presented to eight pilots with a superimposed tracking task. The independent variables included four HUD feature configurations (baseline [no terrain imagery], SVS, EVS, and a combination of SVS and EVS), two visibility conditions, and four legs of the flight. Results indicated that SVS increased overall SA but degraded flight path control performance because of visual confusion with other display features. EVS increased flight path control accuracy but decreased system (aircraft) awareness because of visual distractions. The combination of SVS and EVS generated offsetting effects. Display configurations did not affect pilot spatial awareness. Flight performance was not different among phases of the approach, but levels and types of pilot SA did vary from leg to leg. These results are applicable to development of adaptive HUD features to support pilot performance. They support the use of multidimensional measures of SA for insight on pilot information processing with advanced aviation displays. © 2012 Wiley Periodicals, Inc.     

A shadow detection and extraction algorithm using digital elevation models and x‐band weather radar measurements

The aviation industry has been investigating the potential of synthetic and enhanced vision systems (SVS and EVS) to increase the situational awareness of pilots who are operating in low‐visibility weather conditions. Synthetic vision displays provide a real‐time depiction of a terrain model from the pilot's perspective. To ensure the integrity of this terrain depiction, consistency checking using remote sensing of the terrain environment has been suggested. This requires the detection and extraction of terrain features from both the model and the sensor measurements. Further, the features must be represented in the same reference domain.

Terrain shadowing occurs when areas are not in the line‐of‐sight of the observer. It is these shadowed regions and their morphological characteristics that are identified as the feature domain in which consistency can be assessed between two sources of terrain information. This paper describes an algorithm to extract shadow features from digital elevation models during flight to enable direct comparison with x‐band radar modus operandi. Results are presented using flight‐test data acquired from two platforms with different radar equipment.

The proposed algorithm not only has application to the consistency‐checking problem, but also to terrain navigation, image fusion, and digital elevation model accuracy assessments.     

Statistical Inferences for Generalized Pareto Distribution Based on Interior Penalty Function Algorithm and Bootstrap Methods and Applications in Analyzing Stock Data

This paper studies the application of extreme value statistics (EVS) theory on analysis for stock data, based on interior penalty function algorithm and Bootstrap methods. The generalized Pareto distribution (GPD) models are considered in analyzing the closing price data of Shanghai stock market. The maximum likelihood estimates (MLEs) are obtained by using the interior penalty function algorithm. Correspondingly, the bias and standard errors of MLEs, and the hypothesis test on the shape parameter are concerned through Bootstrap methods. Some simulations are performed to demonstrate the efficacy of parameter estimation and the power of the test. The estimates of the tail index in this paper are compared with those obtained via classical methods. At last, the model is diagnosed by numerical and graphical methods and the Value-at-Risk (VaR) is estimated.     

A neural network accelerator for image analysis

At the heart of this image analysis system are two NET32K analog neural network chips, computing over 100 billion multiply-accumulates per second. The system simultaneously scans sixty-four 16×16-pixel templates over bi-level images, producing feature maps that mark matches between the image and the templates. When we code simple, generic shapes into the templates, the feature maps allow us to make quick, robust analyses of complex, noisy images     

An improved K-means algorithm for underwater image background segmentation

Conventional algorithms fail to obtain satisfactory background segmentation results for underwater images. In this study, an improved K-means algorithm was developed for underwater image background segmentation to address the issue of improper K value determination and minimize the impact of initial centroid position of grayscale image during the gray level quantization of the conventional K-means algorithm. A total of 100 underwater images taken by an underwater robot were sampled to test the aforementioned algorithm in respect of background segmentation validity and time cost. The K value and initial centroid position of grayscale image were optimized. The results were compared to the other three existing algorithms, including the conventional K-means algorithm, the improved Otsu algorithm, and the Canny operator edge extraction method. The experimental results showed that the improved K-means underwater background segmentation algorithm could effectively segment the background of underwater images with a low color cast, low contrast, and blurred edges. Although its cost in time was higher than that of the other three algorithms, it none the less proved more efficient than the time-consuming manual segmentation method. The algorithm proposed in this paper could potentially be used in underwater environments for underwater background segmentation.