一种金属铜破损图像纹理特征提取方法研究

对金属铜破损图像纹理特征点的快速定位,提高破损故障信息分析能力。提出一种基于纹理超像素边缘分割的金属铜破损图像纹理特征提取方法。对采集的金属铜破损图像进行边缘检测和图像融合处理,对降噪提纯后的图像采用自适应纹理密度量化估计方法进行图像的深度超像素特征分割,得到图像的显著性主颜色分量特征信息,计算图像向量量化区域的图像超像素级视觉特征,实现不同像素级深度内的特征匹配和融合,实现图像纹理特征提取优化。仿真结果表明,该算法能准确识别金属铜破损图像纹理特征信息参量,从视觉对比和量化分析角度能实现对破损区域的轮廓分割和检测,性能优越。     

一种基于SIFT的遮挡目标跟踪算法

针对运动目标遮挡的难题,提出一种新的遮挡目标跟踪算法.采用三帧取均值进行背景建模,采用相邻帧差法和背景差分结合自动提取出运动目标,对单运动目标生成SIFI(scale invariant feature transform)向量.当运动目标处于遮挡状态时,将遮挡区域与单运动目标进行SIFT特征匹配,通过特征匹配点的坐标,找出单运动目标在遮挡区域中的位置,并对SIFT特征匹配运用RANSAC算法进行优化,实现遮挡情况下目标的有效跟踪.实验表明,该算法能准确地跟踪处于遮挡中的目标,实现运动目标跟踪的连续性和稳定性.     

一种用于零件定位系统的亚像素边缘检测方法

介绍了基于计算机视觉的长条形杆件测量系统方案,说明在该系统中零件图像边缘提取对测量精度起着重要作用.本文在传统亚像素边缘检测方法基础上,提出垂直占空比亚像素边缘提取方法.该算法具有算法简单、效率高、边缘提取准确的特点,已成功用于长条形杆件测量系统中.     

遮挡情况下目标三维点云的提取研究

针对三维信息提取过程中的遮挡问题,提出利用多基线数字近景摄影测量方法进行多光线的影像匹配,基于遮挡区域约束的三维点云滤波算法,即去除了遮挡目标,又可提取部分被遮挡区域的三维数据点.试验表明,该方法适应了空间分布的不连续性,在遮挡情况下较好地解决了目标的遮挡问题,同时多方向交会使前方交会具有冗余观测,增加了三维数据点的可靠性.     

基于超宽带非对称双边双程测距的矿山井下定位跟踪算法研究

在地下矿山建立稳定、有效、高精度的人员定位系统和算法，对矿井事故灾害的灾时指导逃生和灾后救援等方面起着至关重要的作用。基于超宽带定位测距系统，进行了室内非对称双边双程测距试验，利用粒子群算法、Taylor级数迭代等智能算法完成对目标的定位，以期获得更精确的定位解析，相较于传统的Wi-Fi通信系统及射频识别人员定位管理方法，能够在保证实时性的同时，有效、准确地提供遇险人员的位置及周边地形状况，达到了厘米级定位精度，实现了井下精确定位，为井下救援工作提供技术支持。     

多媒体传感器网络中被动目标定位算法

针对被动目标定位问题,提出了一种多视觉感知节点协同定位算法.与传统的视觉目标定位方法不同,该算法无需进行训练或特征识别,而是各个节点基于机器视觉理论,将目标投影为定位平面上的一条直线,并使用Hough变换将该直线映射为Hough空间中的一个离散点,最后对离散点进行正弦曲线拟合,拟合参数即为目标的最终位置估计.根据视觉感知节点所处高度及分辨率引入置信度参数,进一步提高了多视觉感知节点协同定位算法精度.最后分析了算法的时间复杂度,并通过实验验证了该算法可以获得5 cm以上的定位精度.     

基于平行光背光成像的物体边缘测量方法与试验研究

机器视觉测量技术由于其非接触、效率高和测量一致性好等优点,广泛地应用在自动化大生产中.提出了一种通过平行光背光成像进行物体边缘定位和测量的方法.采用激光扩束器获得平行光光源,该光源穿过物体边缘照射到一个光幕上,利用图像传感器获得光幕上的物体边界的位置并进行量化分析,最终实现边界的定位和测量.对提出的方法进行试验测试和分析,结果表明该方法简单可靠,测量效果稳定.     

一种基于卷积神经网络的CSI指纹室内定位方法

针对提高Wi-Fi指纹室内定位技术性能,提出了一种基于卷积神经网络（Convolutional neural networks,CNN）的信道状态信息（Channel state information,CSI）指纹室内定位方法。在离线阶段联合定位环境参考点的幅度差和相位差信息,利用CNN进行训练,保存训练后的CNN网络模型作为指纹;在线阶段,针对不同实验场景,对测试数据的幅度差信息和相位差信息进行加权处理,引入改进的基于概率的指纹匹配算法,利用待定位点的CSI信息并通过CNN网络模型预测待定位点的坐标。此外,为增强算法普适性,针对复杂室内场景,提出了双节点定位方案来提高定位精度。在廊厅和实验室室内两种不同定位场景进行了实验,信息联合定位算法分别获得了24.7 cm和48.1 cm的平均定位误差,验证了基于CNN的CSI幅度差和相位差联合定位算法的有效性。      

基于FCM-WA联合算法的多种类矿石图像分割

矿石图像分割是基于机器视觉的矿石粒度分布检测的重要组成部分。针对复合矿山中颜色多样、纹理复杂且边缘粘连的多种类矿石图像难以识别与分割的问题，提出了一种基于FCM-WA联合算法的矿石图像分割方法。首先对矿石图像进行形态学优化，利用双边滤波、直方图均衡化和形态学重构来优化矿石图像的几何特征，减少噪声对分割效果的影响，提高图像对比度；然后将模糊C均值聚类（FCM）算法与分水岭（WA）算法相结合，利用FCM算法进行聚类迭代，计算出合适的分割阈值并对矿石图像进行分割，输出二值化图像；再利用基于距离变换的WA算法优化FCM算法的分割结果，对FCM算法输出的矿石图像边缘粘连部分进行分割，以获取最佳的分割图像。研究结果表明：（1）利用形态学优化流程处理矿石图像能够减少噪声并增强边缘信息，从而提高对比度；（2）相比传统的大津法和遗传算法，本文所提FCM-WA方法的稳健性更强、分割效果更好，对多种类的矿石图像像素分割准确率和矿石粒度识别准确率均可达到92%以上；（3）通过试验验证，FCM-WA方法能够精确地分割颜色多样、纹理特征复杂及边缘粘连的多种类矿石图像，分割结果满足粒度分布检测的要求；（4）FCM-...     

基于半监督流形学习的无线传感器网络定位方法及其应用

提出了一种基于流形半监督学习的移动节点定位算法.该算法利用基于流形学习的半监督方法,通过一定量的有标签样本和无标签样本,获取隐含在节点接收信号强度信息中的流形结构,直接建立节点物理位置与接收信号强度之间的映射关系.算法不需要使用现有的理论或经验信号传播模型,避免了模型不准确带来的定位误差,而且允许网络中存在大量无标签样本,降低了数据采集难度,提高了算法实用性.冶金工业现场的实际应用结果表明,相对RADAR算法,本文算法具有较高的定位精度.     

Expression of transforming growth factor-alpha (TGF-alpha) in cholesteatoma

BACKGROUND: This report describes the measurement of convergent eye movements and calculation of the spatial convergence point from the angular eye position data. These measurements were made in the dark while the subject experienced inertial motion aboard a centrifuge. This was an exploratory experiment with the goal of evaluating the eyes' convergence in the dark, and to see if this convergence point is dependent on inertial motion. METHODS: The subject was rotated in the dark on NAMRL's Coriolis Acceleration Platform in Pensacola, FL. The pupil positions were tracked by two helmet-mounted infrared cameras connected to a computer-controlled data acquisition system. We used the position data to calculate the angles through which the eyes rotated, and then applied trigonometric principles to construct the line of sight for each eye for any instant in time. The intersection of these two lines of sight was the convergence point. RESULTS: With the binocular eye-tracking system, we could accurately determine an accelerating subject's convergence point to within 10%, if the point was less than 1.5 m away from the subject. At convergence distances greater than 1.5 m, the angular movements of the two eyes became so small that determining a convergence point was difficult.     

Application of MintDrive Automatic Precision Positioning System

It is very important to locate batteries accurately and quickly during automatic battery production.Unstable or inaccurate location will negatively influence battery's consistency, quality and finished product rate.A traditional way is using sensor to detect and locate batteries directly , but because of the detecting tolerance, setting them on a fixed point exactly is almost impossible.This problem could be completely solved by the application of mint drive automatic accurate servo locating system.Firstly operating software WorkBench test was applied to collocate the servo locating driver for a most optimized control.Then based on the requirement of real location, program and test the locating action with a programming software and finally upload all the locating information to MicroLogix 1200 PLC, the PLC will control the running on each station telling when to locate, where is the location and how to eliminate bad parts.For this intelligent servo locating system has the advantages of powerful function, simple operation, high controlling and locating accuracy and easy maintenance, it is very suitable to be adopted by automatic battery making line.It is regarded as a very advanced method of control currently for reducing waste material due to inaccurate location and tough adjustment.     

基于单目线结构光的智能仓库车辆识别系统

为了实现钢铁企业智能仓库的车辆识别,提出通过基于机器视觉技术的智能仓库车辆识别方案。将面阵CCD相机及线结构光布置在车辆停车位上方拍摄车辆及结构光图像,首先通过张正友标定法及光平面空间原理完成相机与光平面标定;其次使用图像预处理算法及骨架法提取结构光光条中心线;然后通过引入车上、车下点的判别标准,精准识别车上边缘点,并引入车辆高度变量,得到车上边缘点由图像坐标向世界坐标的转化模型;最后根据车上3个边缘点坐标计算车辆位置信息,完成智能库区车辆方位识别。研究结果表明,所构建的测量系统及算法可以获取仓库车辆位置信息,通过标准件精度验证试验可知系统测量结果精确度较高,能够准确实时反映仓库车辆位置信息,这为智能仓库车辆位置检测提供了有效测量手段。     

基于单目线结构光的智能仓库车辆识别系统

为了实现钢铁企业智能仓库的车辆识别,提出通过基于机器视觉技术的智能仓库车辆识别方案。将面阵CCD相机及线结构光布置在车辆停车位上方拍摄车辆及结构光图像,首先通过张正友标定法及光平面空间原理完成相机与光平面标定;其次使用图像预处理算法及骨架法提取结构光光条中心线;然后通过引入车上、车下点的判别标准,精准识别车上边缘点,并引入车辆高度变量,得到车上边缘点由图像坐标向世界坐标的转化模型;最后根据车上3个边缘点坐标计算车辆位置信息,完成智能库区车辆方位识别。研究结果表明,所构建的测量系统及算法可以获取仓库车辆位置信息,通过标准件精度验证试验可知系统测量结果精确度较高,能够准确实时反映仓库车辆位置信息,这为智能仓库车辆位置检测提供了有效测量手段。     

Recording eye movements using coaxial cameras--applications for visual ergonomics and reading studies

We developed a system of coaxial video cameras that records monocular eye position and scene, and superimposes these images using a digital video mixer. We mounted miniature video cameras above and below a cube beam-splitting prism in the spectacle plane. An infrared emitting diode was imaged in the cornea to locate eye position. The technique was accurate to about 0.5 degrees within 15 degrees of primary gaze; however, we see its main advantages as being its low cost and simple design that, for some applications, does not require complex computer analysis and data manipulation. With improved camera optics, it has the potential for helmet mounting and use remote from a recording console. We used the instrument to monitor a reader's eye position when using low vision devices, and see applications of the technique in the field of visual ergonomics and sports vision.     

Locus of interpretive and inference processes during text comprehension: A comparison of gaze durations and word reading times.

Investigated the locus of interpretive and inference processes during text comprehension. Two positions were contrasted: the buffer-integrate-purge position, which assumes that text-level interpretive and inference processes operate at sentence or clause boundaries, and the immediacy position, which assumes that interpretive and inference processes operate as soon as possible. Two methods of collecting word reading times were contrasted: gaze durations and self-paced word reading times. In simple narrative passages, there was an increase in word reading times for end-of-clause words when self-paced reading times were collected, but there was a decrease when gaze durations were measured with eye tracking equipment. The data indicate that interpretive and inference processes operate immediately and that buffering is to some extent an artifact of the self-paced moving window method. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Echocardiographic findings before and after interventional occlusion of a persistent ductus arteriosus Botalli using the Rashkind Occluder System

AIM: The aim of this study was to investigate whether before releasing the Rashkind occluder system for occlusion of persistently patent ductus arteriosi Botalli from the delivery catheter its position can be adequately demonstrated sonographically to avoid angiography. METHOD: From March to August 1993 occlusion of a persistently patent ductus arteriosus by means of the Rashkind system was carried out in 10 patients. A standard technique using transthoracic echocardiography was used. RESULTS: Echocardiography allowed to demonstrate and locate the occluder system in all cases. CONCLUSION: Echocardiography is a reliable means for locating the Rashkind occluder system avoiding angiography.     

Examination of the two-dimensional pupil function in coherent scanning microscopes using spherical particles

The determination of the modulus of the pupil function in reflection acoustic and optical microscopy with the help of spherical particles is demonstrated. The theoretical examination shows that this method can give good results if the pupil function drops smoothly towards the edge. For a pupil with a sharp edge this technique can give the accurate pupil function only by imaging large spheres. The dependence of the accuracy of the method on the size of the spherical particle is analyzed. Numerical and experimental results obtained for the confocal scanning optical microscope and the scanning acoustic microscope are examined.     

Eye position influence on the parieto-occipital area PO (V6) of the macaque monkey

The aim of this work was to study the effect of eye position on the activity of neurons of area PO (V6), a cortical region located in the most posterior part of the superior parietal lobule. Experiments were carried out on three awake macaque monkeys. Animals sat in a primate chair in front of a large screen, and fixated a small spot of light projected in different screen locations while the activity of single neurons was extracellularly recorded. Both visual and non-visual neurons were found. About 48% of visual and 32% of non-visual neurons showed eye position-related activity in total darkness, while in approximately 61% of visual response was modulated by eye position in the orbit. Eye position fields and/or gain fields were different from cell to cell, going from large and quite planar fields up to peak-shaped fields localized in more or less restricted regions of the animal's field of view. The spatial distribution of fixation point locations evoking peak activity in the eye position-sensitive population did not show any evident laterality effect, or significant top/bottom asymmetry. Moreover, the cortical distribution of eye position-sensitive neurons was quite uniform all over the cortical region studied, suggesting the absence of segregation for this property within area PO (V6). In the great majority of visual neurons, the receptive field 'moved' with gaze according to eye displacements, remaining at the same retinotopic coordinates, as is usual for visual neurons. In some cases, the receptive field did not move with gaze, remaining anchored to the same spatial location regardless of eye movements ('real-position cells'). A model is proposed suggesting how eye position-sensitive visual neurons might build up real-position cells in local networks within area PO (V6). The presence in area PO (V6) of real-position cells together with a high percentage of eye position-sensitive neurons, most of them visual in nature, suggests that this cortical area is engaged in the spatial encoding of extrapersonal visual space. Since lesions of the superior parietal lobule in humans produce deficits in visual localization of targets as well as in arm-reaching for them, and taking into account that the monkey's area PO (V6) is reported to be connected with the premotor area 6, we suggest that area PO (V6) supplies the premotor cortex with the visuo-spatial information required for the visual control of arm-reaching movements.     

Are Eyes Windows to a Deceiver's Soul? Children's Use of Another's Eye Gaze Cues in a Deceptive Situation.

Three experiments examined 3- to 5-year-olds' use of eye gaze cues to infer truth in a deceptive situation. Children watched a video of an actor who hid a toy in 1 of 3 cups. In Experiments 1 and 2, the actor claimed ignorance about the toy's location but looked toward 1 of the cups, without (Experiment 1) and with (Experiment 2) head movement. In Experiment 3, the actor provided contradictory verbal and eye gaze clues about the location of the toy. Four- and 5-year-olds correctly used the actor's gaze cues to locate the toy, whereas 3-year-olds failed to do so. Results suggest that by 4 years of age, children begin to understand that eye gaze cues displayed by a deceiver can be informative about the true state of affairs. (PsycINFO Database Record (c) 2010 APA, all rights reserved)