LDE融合光谱回归分类的光照变化人脸识别

针对光照变化人脸识别问题中传统的光谱回归算法不能很好地进行特征提取而严重影响识别性能的问题,提出了局部判别嵌入优化光谱回归分类的人脸识别算法。计算出训练样本的特征向量;借助于数据的近邻和分类关系,利用局部判别嵌入算法构建分类问题所需的嵌入,同时学习每种分类的子流形所需的嵌入;利用光谱回归分类算法计算投影矩阵,并利用最近邻分类器完成人脸的识别。在两大人脸数据库扩展YaleB及CMU PIE上的实验验证了该算法的有效性,实验结果表明,相比其他光谱回归算法,该算法取得了更高的识别率、更好的工作特性,并且降低了计算复杂度。     

Spatial and temporal variability in lake trout diets in Lake Ontario as revealed by stomach contents and stable isotopes

Lake trout (Salvelinus namaycush) are an ecologically and economically important piscivore with reported differences in diet and feeding behaviour throughout its range. Eleven stomach content and stable isotope-based metrics were used to describe diets of 349 lake trout between two years (2013 and 2018) and among geographic zones (west, central, east, Kingston basin) in Lake Ontario. Using individual (e.g., volumetric, %V) and aggregate (e.g., index of relative importance, %IRI) diet metrics, we found an overwhelming dominance of alewife (Alosa pseudoharengus) in lake trout diets among some zones in 2013 (%V = 23.3 – 92.7; %IRI = 12.2 – 99.5) and all zones in 2018 (%V = 83.9 – 96.7; %IRI = 96.5 – 100). Round goby (Neogobius melanostomus) and rainbow smelt (Osmerus mordax) were secondary lake trout prey items with relative diet percentages only marginally reflected by spatial and temporal variation in prey abundance (round goby: %V = 1.0 – 33.3, %IRI = 0.1 – 13.2; rainbow smelt: %V = 2.5 – 54.0, %IRI = 0.1 – 54.0). Carbon (δ¹³C) and nitrogen (δ¹⁵N) isotopic niche areas and orientations were similar across all year-zone combinations reinforcing temporal and spatial consistency in lake trout diet. The findings of this study advance the time series in describing Lake Ontario lake trout diets and can be used to complement stock assessments and management decisions associated with carrying capacity for the diverse salmonid community.     

基于引力搜索算法的分数阶变异时序回归GSA-TSGM(1,1)模型

为了利用分数阶累加算子在灰色短期预测中的高效性能,首次将分数阶累加算子引入变异时序回归模型以期取得更高的预测精度。主要方法如下：首先取湖北省链子崖某监测点1978—1987年的十年数据作为训练集并使用引力搜索算法确定最佳分数阶累加阶数,而1988—1993年的六年数据作为验证集验证提出的模型;其次对比了经典灰色模型GM(1,1)、分数阶累加灰色模型、变异时序回归模型TSGM(1,1)三种灰色模型。结果如下：首先修正了陈西江等人变异时序回归模型仿真时出现的错误,其次表明了相比于其他的模型,基于引力搜索算法的分数阶累加时序回归模型在进行灰色长期预测中具有较高的预测精度。因此,通过分数阶累加算子提高了灰色理论中长期预测模型的精度,为灰色长期预测提供了指导。     

岩石节理三维粗糙度各向异性表征新方法

针对传统二维节理面粗糙度表征方法的不足,提出一种新的岩石节理面三维粗糙度描述方法。首先,利用光学扫描系统得到节理表面点云数据并建立三维模型;运用改进的变异函数方法,提出一种新的基于正态分布模型参数的三维粗糙度(VA)方法;最后,运用新方法对花岗岩张拉型节理表面的粗糙特性进行分析。结果表明,提出的新方法能够反映节理面的起伏变化、频率和分形特征,并因此能描述出节理粗糙度的各向异性特性。     

Face detection based on skin color likelihood

We propose a face detection method based on skin color likelihood via a boosting algorithm which emphasizes skin color information while deemphasizing non-skin color information. A stochastic model is adapted to compute the similarity between a color region and the skin color. Both Haar-like features and Local Binary Pattern (LBP) features are utilized to build a cascaded classifier. The boosted classifier is implemented based on skin color emphasis to localize the face region from a color image. Based on our experiments, the proposed method shows good tolerance to face pose variation and complex background with significant improvements over classical boosting-based classifiers in terms of total error rate performance.     

Critical evaluation of coupling particle size distribution with the shrinking core model

The shrinking core model (SCM) is widely used to model fluid-solid reactions such as the leaching of metals from minerals. In most cases, however, the particle size distribution (PSD) of the solid material was disregarded. In this paper the erroneous shift in the control regime when neglecting PSD was quantified and the dependence of the shift on the coefficients of variation (CV) and the type of PSD was analysed. By coupling the SCM with a Gamma PSD, it was found that neglecting the PSD would shift the control regime from chemical reaction to inert/ash layer diffusion, when the CV was between 0.7 and 1.2. For a system controlled by liquid film diffusion, neglect of the PSD, would shift the control regime to chemical reaction when CV is between 0.3 and 0.7 or to inert/ash layer diffusion when CV is greater (0.9-1.5). It was therefore postulated that some researchers had unknowingly made invalid conclusions about the control regime due to the neglect of PSD. However, an inert/ash layer diffusion-controlled process was insensitive to the neglect of PSD. When CV<0.3, neglect of the PSD would not cause any erroneous shifts, irrespective of the control regime. Experimental data confirmed the observation. For a given CV, the deviation in the fraction reacted from the mono-PSD increases with CV and decreases with time. The maximum deviation, which occurs at the beginning, is about 10% with a gamma PSD of CV=0.3. The percent deviation is dependent of the type of PSDs. Gamma PSD gives the lowest deviation while Gaudin-Schuhmann results in the largest deviation (maxi. ∼19%, with CV=0.3) in the first half of dissolution process. Log-normal distribution gives a larger deviation than gamma but quickly approaches the latter with time. The deviation for Rosin-Rammler is between log-normal and Gaudin-Schuhmann. For systems with CV less than 0.3, the SCM can be fairly used without considering PSD. When CV is greater than 0.3, particularly in the early stage of a dissolution process with a PSD other than gamma, PSD should be included to avoid substantial errors.     

Coupling of dry matter and nitrogen accumulation in ryegrass

The logistic model has proven very useful in relating dry matter production of warm season perennial forage grasses to applied nitrogen. A recent extension of the model coupled dry matter and plant N accumulation through a common response coefficient c. The objective of this analysis was to apply the extended logistic model to cool season Gulf annual ryegrass (Lolium multiflorum Lam.) and to establish a common response coefficient c between accumulation of dry matter and plant N. Analysis of variance established the validity of this hypothesis. The model accurately described response of dry matter, plant N removal, and plant N concentration to applied N, with an overall correlation coefficient of 0.9954. Furthermore, the model closely described the relationship between yield and plant N removal, as well as between plant N concentration and plant N removal. The logistic equation is well-behaved and simple to use on a pocket calculator. It can be used to estimate yields and plant N removal in evaluation of agricultural production and environmental quality.     

昭通横江流域年径流预测

针对昭通横江由南到北贯穿昭通全境,水资源量丰富,但南北差异大,南部少、北部多的特点,根据流域内水文站实测年径流、大气环流指数、海温指数及其他指数等资料系列,采用多元回归分析法拟合1981—2010年径流,建立了预测模型,验证、预测分析了2011—2016年径流量。结果表明:干流控制站预测精度高于干流上段及支流代表站,水利工程建设等人类活动影响是导致出现这样结果的主要原因;多元回归分析在横江流域径流预测中具有一定的适应性,但因指标的差异及相关参数可获取性的差异,预测结果存在区域间、等级间的差别。     

城市径流总量控制指标分解及实现路径探索与实践

针对海绵城市径流总量控制目标,提出一种在城市规划体系中控制性详细规划和修建性详细规划阶段分解落实年径流总量控制率的方法。在控制性详细规划阶段分解径流总量控制指标时,分步骤先后确定各项低影响开发措施的面积率和下沉深度;在修建性详细规划阶段,配合径流总量控制指标分解结果,给出适宜的各项低影响开发措施的技术实施导则,以保证各单项指标的正确实施,并以某项目为例介绍了技术实施导则的构建方法。     

灞河流域气候因子对水沙变化的影响

利用灞河流域蓝田气象站和马渡王水文站1960—2012年的气象、水文实测资料,分析灞河流域气候及水沙变化规律,同时运用相关性分析、灰色关联分析、多元线性回归模型等多种方法探讨了该流域水沙变化与气候变化的关系。结果表明:灞河流域降雨量、蒸发量、径流量和输沙量皆呈显著下降趋势,而气温呈上升趋势;降雨量与水沙都有重要的相关关系,1960—1990年影响径流量的气候因子敏感度由大至小依次为降雨量、气温、蒸发量,而1991—2012年则为降雨量、蒸发量、气温,当气温和蒸发量不变时,降雨量每增加1 mm,两阶段的年径流量分别增加0.14亿m3和0.08亿m3;1960—2012年影响输沙量的气候因子敏感度由大至小依次为降雨量、气温、蒸发量,当气温和蒸发量不变时,降雨量每增加1 mm,年输沙量增加0.668万t。