子空间干扰加高斯杂波背景下基于GLRT的斜对称方向检测器设计

文章研究了背景为子空间干扰加高斯杂波的距离扩展目标方向检测问题。杂波是均值为零协方差矩阵未知但具有斜对称特性的高斯杂波,目标与干扰分别通过具备斜对称特性的目标子空间和干扰子空间描述。针对方向检测问题,利用上述斜对称性,根据广义似然比检验(Generalized Likeli-hood Ratio Test,GLRT)准则的一步与两步设计方法,设计了基于GLRT的一步法与两步法的距离扩展目标方向检测器。通过理论推导证明了这2种检测器相对于未知杂波协方差矩阵都具有恒虚警率。对比相同背景下已有检测器,特别是在辅助数据有限的场景下,文章提出的2个检测器表现出了优越的检测性能。

基于多方向滤波和邻域去噪的掌纹主线提取

为了提取完整的、接近自然生长规律的掌纹主线,提出了一种基于多方向滤波和邻域去噪的掌纹主线提取方法.首先对全手掌图像进行预处理,得到统一大小和灰度的ROI(Region of Interest,ROI)掌纹图;然后用中值和4个方向检测器对ROI图像滤波,并对滤波后的结果进行底帽运算从而得到粗提取的结果;最后对粗提取的主线进行二值处理和邻域去噪,得到最终的主线提取结果.在公开的数据集上将新方法与现有同类方法进行对比,主观实验结果表明:新方法提取得到的主线图像连续且完整,客观的PSNR值比对比文献平均高约2.3153 dB,表明新方法能较好地去除断线、孤点噪声,从而完整、有效地提取掌纹主线.

空间真空测量技术研究

针对国内外在空间真空测量方面的研究状况作了简要描述,对研制的2种空间真空规(转换器方向性真空规和束检测器方向性真空规)进行了介绍,并对空间真空测量技术的后续研究工作作了展望。

A new approach for salt dome detection using a 3D multidirectional edge detector

Accurate salt dome detection from 3D seismic data is crucial to different seismic data analysis applications. We present a new edge based approach for salt dome detection in migrated 3D seismic data. The proposed algorithm overcomes the drawbacks of existing edge-based techniques which only consider edges in the x （crossline） and y （inline） directions in 2D data and the x （crossline）, y （inline）, and z （time） directions in 3D data. The algorithm works by combining 3D gradient maps computed along diagonal directions and those computed in x, y, and z directions to accurately detect the boundaries of salt regions. The combination of x, y, and z directions and diagonal edges ensures that the proposed algorithm works well even if the dips along the salt boundary are represented only by weak reflectors. Contrary to other edge and texture based salt dome detection techniques, the proposed algorithm is independent of the amplitude variations in seismic data. We tested the proposed algorithm on the publicly available Netherlands offshore F3 block. The results suggest that the proposed algorithm can detect salt bodies with high accuracy than existing gradient based and texture-based techniques when used separately. More importantly, the proposed approach is shown to be computationally efficient allowing for real time implementation and deployment.

基于多部位多示例学习的人体检测

基于部位的检测方法能处理多姿态及部分遮挡的人体检测,多示例学习能有效处理图像的多义性,被广泛应用于图像检索与场景理解中.文中提出一种基于多示例学习的多部位人体检测方法.首先,根据人体生理结构将图像分割成若干区域,每个区域包含多个示例,利用AdaBoost多示例学习算法来训练部位检测器.然后利用各部位检测器对训练样本进行测试得到其响应值,从而将训练样本转化为部位响应值组成的特征向量.再用SVM方法对这些向量进行学习,得到最终的部位组合分类器.在INRIA数据集上的实验结果表明该方法能改进单示例学习的检测性能,同时评价3种不同的部位划分及其对检测性能的影响.

A novel orientation measurement using optical sensor for spherical motor

The measurement of spherical rotor orientation is crucial to the close-loop control of spherical motors. This paper presents a novel method for the measuring of three-degree-of-freedom (DOF) rotor orientation of spherical motors using optical sensors. The spatial orientation of spherical rotor is output in the form of ZXZ Euler angles. Firstly, the structure of the measuring system composed of optical sensors and the patterns on the rotor surface are presented, and the operational principle of recognizing intersection points between the optical ring detectors and the latitude/longitude on the rotor surface is illustrated. The analytical model of input-output characteristic is established for the measuring system of three-DOF rotor orientation. Afterwards, the effect of parameters of the optical ring detectors on the linearity, sensitivity, resolving power and measuring range of the measuring system is analyzed using the analytical model. Finally, the feasibility of the measurement is validated through experiments of prototype measuring system. The analysis is expected to be a basis for the design parameter optimization of the orientation measuring system of a PM spherical motor.