归一化径向基函数网络的结构优化策略

针对归一化RBF网络的结构设计，首先利用基于山峰函数的减法聚类算法构造网络的初始结构和初始数值，并采用奇异值分解（SVD）算法分析了网络初始结构中隐含层节点与奇异值，累积贡献率以及索引向量的关系，提出了针对归一化RBF网络的结构精简优化策略。最后，对该结构优化策略的可行性和有效性进行了仿真验证和性能比较。     

基于多阶抽样谱图聚类彩色图像分割

针对谱聚类应用于图像分割时权矩阵的谱难以计算的实际问题,设计了一个图像多阶抽样谱图聚类算法.首先,给出了采样数定理及其证明,并推导出与聚类类别数和最小聚类数相关的最小采样数目;其次,根据最小采样数数目,对像素点进行均匀采样,并利用谱聚类对采样点进行聚类,设计一个罚函数,通过多次抽样,消除抽样对谱聚类模型稳定性的影响;最后,定义了像素点和类之间的距离,对剩余的点按距离最近原则进行聚类.实验结果表明了算法的有效性.     

The temporal dimension of differenced Normalized Burn Ratio (dNBR) fire/burn severity studies: The case of the large 2007 Peloponnese wildfires in Greece

The temporal dimension of differenced Normalized Burn Ratio (dNBR) fire/burn severity studies was studied for the case of the large 2007 Peloponnese wildfires in Greece. Fire severity is defined as the degree of environmental change as measured immediately post-fire, whereas burn severity combines the direct fire impact and ecosystems responses. Geo Composite Burn Index (GeoCBI), two pre-/post-fire differenced Thematic Mapper (TM) dNBR assessments and a Moderate Resolution Imaging Spectroradiometer (MODIS) dNBR time series were used to analyze the temporal dimension. MODIS dNBR time series were calculated based on the difference between the NBR of the burned and control pixels, which were retrieved using time series similarity of a pre-fire year. The analysis incorporated the optimality statistic, which evaluates index performance based on displacements in the mid-infrared-near infrared bi-spectral space. Results showed a higher correlation between field and TM data early post-fire (R² = 0.72) than one-year post-fire (R² = 0.56). Additionally, mean dNBR (0.56 vs. 0.29), the dNBR standard deviation (0.29 vs. 0.19) and mean optimality (0.65 vs. 0.47) were clearly higher for the initial assessment than for the extended assessment. This is due to regenerative processes that obscured first-order fire effects impacting the suitability of the dNBR to assess burn severity in this case study. This demonstrates the importance of the lag timing, i.e. time since fire, of an assessment, especially in a quickly recovering Mediterranean ecosystem. The MODIS time series was used to study intra-annual changes in index performance. The seasonal timing of an assessment highly impacts what is actually measured. This seasonality affected both the greenness of herbaceous resprouters and the productivity of the control pixels, which is land cover specific. Appropriate seasonal timing of an assessment is therefore of paramount importance to anticipate false trends (e.g. caused by senescence). Although these findings are case study specific, it can be expected that similar temporal constraints affect assessments in other ecoregions. Therefore, within the limitations of available Landsat imagery, caution is recommended for the temporal dimension when assessing post-fire effects. This is crucial, especially for studies that aim to evaluate trends in fire/burn severity across space and time. Also, clarification in associated terminology is suggested.     

A novel global harmony search algorithm for task assignment problem

The objective of task assignment problem (TAP) is to minimize the sum of interprocessor communication and task processing costs for a distributed system which subjects to several resource constraints. We use a novel global harmony search algorithm (NGHS) to solve this problem, and the NGHS algorithm has demonstrated higher efficiency than the improved harmony search algorithm (IHS) on finding the near optimal task assignment. We also devise a new method called normalized penalty function method to tradeo^® the costs and the constraints. A large number of experiments show that our algorithm performs well on finding the near optimal task assignment, and it is a viable approach for the task assignment problem.     

Seated body apparent mass response to vertical whole body vibration: Gender and anthropometric effects

The gender and anthropometric effects on apparent mass characteristics of the seated body exposed to vertical vibration are investigated through laboratory measurements. The study was conducted on 31 male and 27 female subjects, exposed to three levels of vertical vibration (0.25, 0.50 and 0.75 m/s² rms acceleration) in the 0.50 to 20 frequency range, while seated without a back support and against a vertical back support. The apparent mass responses were analyzed by grouping datasets in three ranges of mass-, build- and stature-related parameters for the male and female subjects. Comparisons of responses of male and female subjects with comparable anthropometric properties showed distinctly different biodynamic responses of the two genders. The primary resonance frequency of male subjects was significantly (p < 0.001) higher than the female subjects of comparable body mass but the peak magnitude was comparable for both the gender groups. The male subjects showed greater softening with increasing excitation magnitude compared to the female subjects, irrespective of the sitting condition. The male subjects showed significantly higher peak magnitude response than those of the female subjects for the same anthropometric properties, except for the total and lean body mass. The peak magnitude was linearly correlated with the body mass, body mass index, body fat and hip circumference (r² > 0.7), irrespective of the back support and excitation conditions for both the genders.Relevance to the industryThe apparent mass responses of the human body exposed to whole-body vibration form an essential basis for an understanding of mechanical-equivalent properties of the body, developments in frequency-weightings for assessment of exposure risks and anthropodynamic manikins for assessment of seats. The effects of gender and anthropometric parameters on the AM response are vital for seeking better seat designs, and anthropodynamic manikins for assessments of seating for male as well as female workers.     

基于方差归一化失真测度的改进的LBG算法

矢量量化（ＶＱ）技术在话者识别系统中得到了广泛的应用。 ＶＱ码本的产生通常采用 ＬＢＧ算法,失真测度则为对矢量的各分量等权重的欧氏距离。在话者识别系统中特征矢量的各个分量的分布是有差别的,且对于不同的话者,这种差别的程度又是不一样的。由于不同分布的各维参数对话者识别的有效性各不相同,因此,文章提出了一种能反映这种有效性差别的失真测度,即：方差归一化失真测度。以该失真测度为基础,并结合时序相关的初始码本设计方法及有效的零胞腔处理技术,文章提出了改进的ＬＢＧ算法,同时利用该算法训练出改进的ＶＱ话者模型,并进行了话者识别实验。     

基于神经网络的边缘强度互相关匹配可信度分析

匹配可信度是分析图象匹配质量的主要指标,针对归一化边缘强度互相关匹配算法,研究了基于神经网络的匹配可信度判别方法,即以参考图与若干实时图的匹配实验结果作为训练用样本数据,然后利用ＢＰ网络进行训练,再将训练后的网络用于匹配可信度的判别。通过实际卫片与航片图象对的匹配实验,证实了该方法的有效性。     

Forecasting warranty performance in the presence of the ‘maturing data’ phenomenon

Forecasting of warranty performance helps car engineers to fine-tune their strategies for warranty cost reduction. The forecasting process involves prediction of not only the specific months-in-service (MIS) warranty performance at a certain future time, but also future MIS values. However, the ‘maturing data’ phenomenon that causes a warranty performance measure at specific MIS values to change with time make such forecasting challenging. Although dynamic linear models have been used for forecasting warranty performance, the focus mainly has been to utilize previous-model-year vehicle data for the analysis. In this paper, we apply a neural network model to forecast year-end warranty performance in the presence of the ‘maturing data’ phenomenon. We use a special type of neural network, viz. radial basis function (RBF), and optimize its parameters by minimizing training and testing errors through planned experimentation. This application shows the effectiveness of RBF neural networks to forecast warranty performance in the presence of the ‘maturing data’ phenomenon.     

基于模板匹配的苹果果实识别

为了实现果蔬收获机器人的果实识别和定位,同时为了消除由于果实大小和生长角度对识别的影响,以自然环境下的果园苹果作为识别对象,采用模板匹配方法,计算目标图与模板图的归一化互相关系数矩阵,通过在空域对模板图像做尺度变换和角度旋转进而对目标图像进行模板匹配,实现对果园苹果的果实识别。实验结果表明,该模板匹配算法能有效识别自然环境下的苹果果实,可以用于机器人视觉系统的识别任务。