基于并查集和边缘检测模板的非理想虹膜定位

非理想虹膜图像往往存在虹膜边缘模糊、灰度变化不均匀、位置偏移及光斑干扰等问题, 这些问题的存在会在一定程度上影响虹膜内、外边界定位的准确率. 针对这个问题, 本文提出采用并查集和边缘检测模板的方法来对非理想虹膜进行内、外边界定位. 内边界定位是首先采用并查集方法完成瞳孔区域粗定位, 然后采用Hough变换对瞳孔进行精确定位; 外边界定位是先利用一系列边缘检测模板大致确定外圆的位置, 然后再依据外边界附近圆环内边缘点的密度来最终完成外边界的精确定位. 实验结果表明, 对于非理想虹膜图像, 本文方法的定位正确率和定位速度均高于其他同类方法.     

Paramedics on the job: Dynamic trunk motion assessment at the workplace

Many paramedics' work accidents are related to physical aspects of the job, and the most affected body part is the low back. This study documents the trunk motion exposure of paramedics on the job. Nine paramedics were observed over 12 shifts (120 h). Trunk postures were recorded with the computer-assisted CUELA measurement system worn on the back like a knapsack. Average duration of an emergency call was 23.5 min. Sagittal trunk flexion of >40° and twisting rotation of >24° were observed in 21% and 17% of time-sampled postures. Medical care on the scene (44% of total time) involved prolonged flexed and twisted postures (∼10 s). The highest extreme sagittal trunk flexion (63°) and twisting rotation (40°) were observed during lifting activities, which lasted 2% of the total time. Paramedics adopted trunk motions that may significantly increase the risk of low back disorders during medical care and patient-handling activities.     

A novel approach for fit analysis of thermal protective clothing using three-dimensional body scanning

The garment fit played an important role in protective performance, comfort and mobility. The purpose of this study is to quantify the air gap to quantitatively characterize a three-dimensional (3-D) garment fit using a 3-D body scanning technique. A method for processing of scanned data was developed to investigate the air gap size and distribution between the clothing and human body. The mesh model formed from nude and clothed body was aligned, superimposed and sectioned using Rapidform software. The air gap size and distribution over the body surface were analyzed. The total air volume was also calculated. The effects of fabric properties and garment size on air gap distribution were explored. The results indicated that average air gap of the fit clothing was around 25–30 mm and the overall air gap distribution was similar. The air gap was unevenly distributed over the body and it was strongly associated with the body parts, fabric properties and garment size. The research will help understand the overall clothing fit and its association with protection, thermal and movement comfort, and provide guidelines for clothing engineers to improve thermal performance and reduce physiological burden.     

Chemical properties of femoral gland secretions in the desert iguana,Dipsosaurus dorsalis

This study investigates the chemistry of femoral gland secretions in the desert iguana,Dipsosaurus dorsalis (Lacertilia: Iguanidae), and discusses their possible functional significance. Electrophoretic and proton NMR studies indicated that the secretions are composed of approximately 80 % protein and 20% lipid material. Individual differences in polyacrylamide gel banding patterns of femoral gland proteins were found. Reflectance spectroscopy revealed that the secretions strongly absorb longwave ultraviolet light, a feature that may be important in the localization of secretion deposits in the environment.     

基于二元多跳WSN的信道容错的目标定位方法

针对信道退化的多跳无线传感器网络下的定位问题,基于最大似然估计提出一种新的信道容错的定位算法.传感器节点接收到的源信号强度数据被压缩量化为二元信号,经Rayleigh退化信道多跳中继到达融合中心.利用信道退化统计量和译码策略,推导出二元观测数据的似然函数,基于最大似然估计获得目标定位,进一步给出目标位置估计的克拉美-罗下界.仿真结果表明了所提出算法的有效性.     

一种基于SBL和APIT的混合定位算法

由于子区域分割粒度的限制,基于阶次序列的定位算法(Sequence-based localization,SBL)存在边缘区域节点定位误差较大和不能保证平均定位误差性能的问题。针对这些问题,提出了一种基于SBL和APIT的混合定位算法,利用APIT信标三角形切割SBL算法子区域,减小子区域面积,实现了分割粒度的细化。通过预先进行系统训练,优化了混合算法的加权系数,进一步提升了算法的定位精度。仿真结果表明,相比于原算法,所提出的混合算法有效地提升了边界区域节点的定位精度,其平均定位误差降低了17.9%,使基于阶次序列的定位算法的定位精度得到了有效改善。     

EVALUATION OF A METHOD FOR MEASURING MIXING TIME USING NUMERICAL SIMULATION AND EXPERIMENTAL DATA

A method for measuring mixing performance in large-scale vessels is investigated. The method is evaluated experimentally in a pilotscale reactor and by using a model based on the three-dimensional calculation of the flow field. The method is based on injection of a tracer and measuring the concentration with several detectors at different locations. The mixing performance is measured as the decay rate of the difference between the concentration at the detectors and the final concentration.

The model predicts a strong dependence on injection point. The assumption of a first-order decay rate of the concentration deviation is more appropriate for some injection points.     

Three-dimensional simulation using fixed coarse-grid thermal-fluid scheme and conduction heat transfer scheme in distinct element method

Three-dimensional simulation using particle-thermal-fluid coupling scheme with a mixed Lagrangian-Eulerian approach is presented and the scheme is described by extending author's previous research [Y. Shimizu, 2004. “Fluid Coupling in PFC2D and PFC3D”, numerical modeling in micromechanics via particle methods—2004, Proceeding of 2nd International PFC Conference, Balkema, 2004, pp. 281-287]. The scheme solves the continuity, Navier-Stokes (N-S) and thermal energy equations numerically in Eulerian Cartesian coordinates, then derives pressure, velocity vectors and temperature for each fixed cell by considering particle existence. On the other hand, driving forces and thermal energy from fluid are applied to the particles as external body forces and a source term in the force-displacement law and the heat transfer calculations, respectively. This paper describes the thermal scheme for both fluid and particles. Results of two- and three-dimensional simulations involving forced convective heat transfer by fluid and conduction heat transfer within particles are presented.     

基于支持向量机的无线传感器网络节点定位算法

机器学习是利用经验来改善自身性能的一种学习方法,而支持向量机(support vector machine,SVM)作为机器学习中的一种新模式,在解决小样本、非线性及高维模式识别等方面有着其特有的优势.基于支持向量机的节点定位算法利用机器学习算法的特性,实现无线传感网络节点定位.其基本思路是将网络区域划分为若干个等分的小格,每一小格代表机器学习算法中一个确定的类别,机器学习算法在学习了已知的信标节点对应的类别后,对未知节点所处位置进行分类,从而进一步确定未知节点的位置坐标.仿真实验表明,"一对一"节点定位算法有较高的定位精度,对测距误差的容忍性较好,同时对信标节点的比例要求并不高,比较适合用于信标节点稀疏的网络环境中;而"决策树"节点定位算法受覆盖空洞的影响并不大,比较适合应用于节点分布不均匀或者存在覆盖空洞的网络环境中.