香料植物桂花野生种群分布格局的分形分析

采用分形理论中的计盒维数、信息维数和关联维数对长汀县石峰寨桂花野生种群分布格局的多尺度分布规律进行比较分析。结果表明,桂花野生种群的分布格局具有分形特征,其计盒维数为0.8622～1.0058,信息维数为0.8670～0.9849,关联维数为1.0037～1.0757。桂花种群的计盒维数、信息维数、关联维数相差不大,表明该种群的分布格局强度相对复杂,局域空间占据较强,具有集聚分布的趋势。     

光强倒数色度空间的彩色人脸光照预处理

为了解决彩色人脸识别中色彩信息易受光照影响的问题,提出一种基于光强倒数色度空间(IICS)的彩色人脸图像预处理方法。本方法首先将图像均匀地分割成子块;将每个图像块变换为IICS空间中的一个二维数据集,并根据数据集的线性分布特性估计图像块的光照颜色;然后对全部图像块的光照估计进行颜色直方图统计,根据直方图对分块估计的结果进行合并;最后,利用估计得到的光照和对角模型将图像光照校正到标准白光下,用于人脸识别。在AR和FERET人脸库上的实验表明,通过引入本光照预处理,有效增强了彩色人脸识别方法对光照变化的鲁棒性,提高了识别精度。     

林业专题图数字水印版权保护研究

针对林业专题图制作过程复杂、成本高、迫切需要进行版权保护的问题,分析了JPEG图像压缩的不变属性,在此基础上提出了一种DCT域数字水印算法,该算法将水印信息嵌入到JPEG格式的林业专题图像的中高频系数当中。结果表明：因为水印信息是嵌入在中高频系数上,所以不仅嵌入的水印信息不可感知性良好,而且水印的嵌入容量也有所增加;即使对加密林业专题图进行有损压缩时使用高于预设值的质量因子,此算法仍然能够完全一致地提取出嵌入的水印信息,并且对各种噪声干扰都具有较好的鲁棒性。     

毛竹种群空间格局的地统计学分析

为了增加对竹林生态系统认识,应用地统计学中的半方差、半方差理论模型模拟、分形维数及其与竹林结构部分因子相关分析,研究毛竹种群空间格局特征。研究结果:(1)毛竹种群半方差总体均值为0.818,林地之间变异较大,属中等强度变异。半方差的标准差较小,占半方差均值的1.293%～7.241%。不同类型毛竹林半方差均值比较两两之间差异不显著。半方差的理论模型模拟多为线性模型,而球状模型和指数模型较少,毛竹林空间自相关性较弱。(2)毛竹林分形维数均趋近于2,表明毛竹林具有较好的均质性。(3)基台值与立竹度、均匀度呈显著相关关系,而分形维数与立竹度和均匀度相关不显著,表明了基台值与分形维数是刻画种群空间格局两方面指标。结论:毛竹林是趋于同质性植被,空间自相关性较弱,毛竹种群半方差在不同类型竹林之间差异不显著;竹林经营时采用竹阔混交林为宜,同时使毛竹在空间分布上趋于均匀,减小半方差,减弱毛竹之间的空间资源竞争强度。     

PAM对潮土水稳性团聚体的影响

通过室内土柱培养,研究四种不同浓度PAM对潮土水稳性团聚体的改良效果。结果表明,加入不同浓度PAM均可促使>0.25mm水稳性团聚体的形成,并有效降低团聚体分形维数,改善土壤结构。随着PAM浓度的增加,改良效果逐渐增强。在加入浓度为1g/kg时达到最好的效果,较对照,>0.25mm水稳性团聚体增加了42.69%,分形维数降低了10.11%。关键词：聚丙烯酰胺（PAM）;水稳性团聚体;分形维数     

基于PSO-SVM的高光谱数据降维的可靠性研究

PSO结合SVM算法对高光谱数据波段进行优化,每次搜索结果不一定相同,因此很多学者对此类算法的可靠性存在疑问。为了证明PSO-SVM降维算法的可靠性,利用PSO-SVM算法对杉木和马尾松的幼中成熟林的高光谱原始数据、一阶微分变换数据、对数变换数据及归一化变换数据进行降维运算,对降维后选择的波段分别利用支持向量机（SVM）、BP神经网络、Mahalanobis距离分类法、Fisher分类法及贝叶斯分类法进行分类,分类结果中,Fisher分类法的结果最好,所有的分类结果均在90%以上,SVM和BP神经网络的分类结果都保持在80%以上,贝叶斯分类法分类精度最差,所有分类结果均未超过90%,最差结果为43.75%。同时,将PSO-SVM与PCA算法进行对比分析,发现在马尾松和杉木的分类过程中PSO-SVM算法优于PCA算法。最后得出结论,PSO-SVM算法提取的特征对Fisher、SVM及BP神经网络分类法有效;当光谱数据差异非常微小时,PSO-SVM比PCA对特征的提取更有效。     

QTL作图中零假设检验统计量分布特征及LOD临界值估计方法研究

研究QTL作图的零假设检验统计量分布特征, 可以帮助我们选取合适的LOD临界值, 以控制全基因组第一类错误的概率。本文利用模拟方法, 研究了QTL作图中单个扫描位点的似然比检验(LRT)统计量在零假设下的分布特征、影响最大LOD统计量累积分布的因素以及不同群体在不同标记密度下有效独立检验次数与染色体长度的关系。结果表明, 在定位加显性效应QTL的一维扫描和定位上位性互作QTL的二维扫描中, 单个扫描位置上的LRT统计量均服从卡方分布, 其自由度等于检测QTL遗传参数的个数; 染色体个数、群体大小和表型测量误差对零假设下检验统计量的分布没有影响, 即不影响LOD临界值的选取, 而群体类型、标记密度和染色体长度有明显影响, BC₁、RIL和F₂三种类型的群体中, BC₁群体的临界值最小, F₂群体的临界值最大, 标记越密、染色体越长, 对应的LOD临界值越大; QTL一维扫描中有效独立检验次数与染色体长度呈正比, 二维扫描中有效独立检验次数与染色体长度呈二次幂关系。借助Bonferroni矫正, 给出了全基因组显著性水平与单个扫描位点显著性水平间的关系, 因此, 研究者可根据作图群体的群体类型、标记密度和基因组长度, 很方便地确定特定全局显著性概率水平下的LOD临界值。     

河流底泥粒径分形维数与重金属含量相关性——以海南岛南渡江塘柳塘为例

为了揭示河流底泥重金属含量与粒径分形维数、有机质之间的数量关系,借助统计分析和土壤分形维数分析法,分析了南渡江塘柳塘52个底泥样品的粒径分形维数特征;开展了底泥重金属含量与有机质、粒径分形维数的相关性分析、主成分分析及模型回归等。研究结果表明：所调查河流底泥粒径分形维数在2.453~2.971之间,不同粒径分级的分形维数平均值不同;重金属含量与有机质含量、极细砂分形维数、细砂分形维数呈显著的正相关,与粘粒、粉粒分形维数呈负相关,底泥中Zn、Hg受粒径的影响较小。底泥中有机质与粘粒分形维数可作为表征底泥基本理化形状的关键指标。底泥重金属含量（Zn、Hg除外）与有机质存在一元线性数量化关系;底泥重金属（Zn、Hg除外）含量与有机质、粘粒分形维数之间存在二元回归模型的数量化关系,上述模型间的数量化关系对于研究底泥的理化性状有着重要的意义。     

2010-2012年安徽省闪电特征分析与应用研究

为减少因闪电带来的人员生命和财产损失,研究安徽省闪电时空分布特征及其应用。采用Excel、Origin和Surfer软件作为统计分析方法,对2010年3月—2012年12月安徽省逐日闪电数据进行分析。结果表明：安徽地区以负闪为主;闪电月变化特征表现为单峰单谷型,暖季多冷季少,闪电次数集中发生在盛夏7月,6—9月为闪电集中发生时段;日变化上,正、负闪的2个高峰值时段分别出现在01—03时和15—17时,14—18时是一日当中对流性天气最集中时段;闪电强度变化特征表现为暖季高冷季低,月平均闪电强度46.2 kA,正、负闪电月平均强度峰值分别出现在7月和6月,高于30 kA的正、负闪电出现概率走势一致,正、负闪电强度值主要集中在0~30 kA之间;空间分布上,年平均闪电强度和密度高值区多发生在平原、丘陵和山区交接地带,也是安徽地区雷电灾害防御重点区域。同时,利用闪电资料和雷暴日数的关系,得出适用于安徽省16个地市的地闪密度计算公式。研究结果对指导本地工农业生产、工程建设具有非常重要的实际意义。     

衣分不同陆地棉品种的产量及产量构成因素的遗传分析

选用衣分不同的陆地棉品种配置组合,率先将主基因-多基因联合世代分析与双列杂交试验分析相结合,分别从单个和整体基因水平上对棉花产量及产量构成因素进行了遗传研究。对2个高×低衣分组合的主基因-多基因6世代联合分析结果表明,各产量性状至少在1个组合中检测到主基因的存在,说明产量性状主基因存在的普遍性。由2个组合各产量性状的主基因、多基因遗传率比较得出,产量性状的主基因遗传率比多基因遗传率在不同组合间趋势变化相对较稳定;各性状在2个组合中的主基因、多基因遗传率分量不完全相同。衣分、铃重和籽指在2个组合中分别以主基因遗传为主和以多基因遗传为主;子棉产量和皮棉产量在2个组合中均以主基因遗传为主;衣指在组合I中以多基因遗传为主,在组合II中属于典型的多基因遗传;单株铃数在组合I中属于典型的主基因遗传,在组合II中以多基因遗传为主。双列杂交结果表明,陆地棉产量及产量构成因素都有较高的遗传主效应方差,产量性状受加性效应和显性效应共同控制,其中,衣分、衣指以加性效应为主;子棉产量、铃重和籽指以显性效应为主;皮棉产量和单株铃数以加性和显性效应为主。衣分和衣指的普通广义遗传率和普通狭义遗传率均最高,与联合世代分析两性状的总遗传率平均值结果趋势一致。相关和通径分析一致表明,产量构成因素中单株铃数对皮棉产量的贡献最大,衣分次之,铃重最小。     

Auto Mesh Algorithm of the Finite Cover System in the Numerical Manifold Method

A auto mesh algorithm of the finite cover system is presented by abstracting the finite cover system of numerical manifold method as independent data classes.The programs are made by Object Oriented Programming(OOP).The accuracy manifold element can be generated automatically when the continuous material is cut by crack.The calculation result shows that the algorithm is reliable and effective.     

Face recognition method with local matching

From human cognition, a face recognition method with local matching based on statistical learning is proposed. The image is divided into several subimages and each subimage is considered as a weak classifier. The Adaboost learning algorithm is used to train the weak classifiers and construct a strong classifier. As a result, all subimages are effectively combined together to explore the best discriminating power and improve the classification accuracy. Compared with the holistic matching methods, the local matching method is robust to variations in illumination, expression, and pose, etc. The experimental results show that the proposed method can improve the face recognition accuracy and is robust to variations in illumination and expression.     

Single sample face recognition based on B-spline and image gradient

A single sample face recognition algorithm based on B-spline and image gradient is proposed. Image gradient method for face recognition has advantage of illumination invariant. But the recognition rate will be greatly decreased when the image contains noise which will seriously influence gradient information. Traditional methods to reduce noise smooth image at the same time and image gradient reorganization rate will be reduced. As the B-spline filter has the feature which can adjust the order, B-spline filter with different orders can be selected according to the image noise value to minimize noise while preserve image gradient information. Experiments prove that using B-spline and image gradient algorithm can achieve a better recognition rate than traditional filtering method on single sample face recognition problem.     

Anomaly detection based on the multi phase clustering and naive bayes

Anomaly detection method was used for calibration data concentration significantly different from other data objects. In this paper, the multi phase clustering aimed at resolving the import of noise data and the lack of the attributive sample, and changing the traditional passive learning of bayes for active learning ways to structure the superior performance classifier. In the pre processing stage, a clustering algorithm based on density is introduced to handle noise data. And the output of the density based clustering algorithm can be used as the input of K means, which responsible for handling the training samples with absent values. At classification time, we introduce adaboost algorithm into naive bayes to generate a more effective classifier.     

Efficiency of bunch-planted and single-planted seedlings for selecting superior families in sugarcane

Summary The efficiency of family selection in bunch-planted sugareane seedlings was investigated by planting 40 families (20 proven and 20 experimental) as bunches and as single seedlings in adjoining replicated trials. A replicated trial planted the following year. using stem cuttings from random seedlings of each family. provided an independant estimate of the value of each family.Five methods were used to compare bunch and single planting, namely theoretical and realized gains from selection, correlations between the seedling and evaluation trials, discrimination based on proven cross classes, and discrimination based on production of advanced selections. Overall, bunch planting was inferior to single planting for family selection. It was promising for only one method of evaluation (proven cross classes) but this, combined with the relatively small area required for bunch-planted seedlings, indicates that family selection may be possible in such plantings. The use of a number of methods of evaluation showed that incorrect conclusions were likely if only one method were used, and revealed deficiencies in methods which would otherwise be considered satisfactory. The project provided a basis for the design of more critical experiments.The components of weight of cane, namely number of stalks and weight per stalk, are negatively correlated, and showed different relationships with yield in the seedling and evaluation trials. Weight per stalk was an important component in seedling trials, with number of stalks by far the main component in the evaluation trial which was planted by the commerical method using stem cuttings.Family selection based on number of stalks and sugar content, preceding mass selection, is likely to provide a better selection system than the usual mass selection in bunch-planted seedlings.Families do not express their full genetic potential in original seedlings. whether bunch or single planted: trials planted from stem cuttings give more effective family selection. Furthermore, family selection for the most important characters (net merit grade and weight of sugar) was much more effective in ratoon than in plant crops. These results have important implications for the practical selection program, further experimental work being necessary to specify the most efficient system.     

Rectangular Cover System of Manifold Method and Its Auto Mesh Algorithm

In contrast to the finite element method, another form of numerical method which is based on the using of rectangular math covers in manifold method is proposed. In this method, the element data is not a necessary in the input data files and it can be incorporated with the CAD technology conveniently. The theory and application of the method are discussed and the auto mesh algorithm of the cover system is also given in this paper. The numerical examples indicate that the method and the algorithm are accurate and effective.     

Temporal stability of multivariate genetic divergence in cacao ( Theobroma cacao L.) in Southern Bahia conditions

Stability of the genetic divergence among five non-commercial cacao cultivars at advanced ages was investigated over a five-year period (1986–1990). Cluster analysis was performed on five yield components measured on harvests from each crop year and on the data pooled over five years. The temporal stability was assessed by using clustering procedure, which involved the calculation Mahalanobis distance (D²) and of Tocher's algorithm applied to the distance matrix. The comparison of D² and of clusters based upon pooled analysis, and taken as the standard, with D² values and clusters obtained from each year showed a stable clustering pattern in the most favourable year. For the rest of the years, inconsistency in number and composition of clusters formed was observed. An optimum environment was shown to be important for better expressing divergence by D². Consequently, the evaluation of the genetic divergence among the studied cultivars may be conducted based on a single favourable year. This will result in considerable savings in man power, financial resources, and time which would be wasted should the evaluation be extended for several years.     

Prediction of winter wheat cultivar performance in Germany: At national,regional and location scale

Winter wheat cultivar recommendation is usually based on the cultivar performance observed in post-registration trials. In Germany, official recommendations are based on state cultivar trials, which are conducted individually by the federal states, usually over a period of three years. In each predefined winter wheat cultivation region a subset of registered cultivars is tested. The recommendation in a particular region is mainly based on the yields from trials on several locations in this region. Practically, the farmer's interest is a prediction of the yielding ability of cultivars on his own farm in the following growing season. This prediction can be made based on data from different scales, and with one year or multiple-year data. Here, we evaluated the prediction ability with the data from national, regional and location scales per se, and tried to find the optimal information source (scale and number of years) to predict the relative yield of a specific cultivar for a specific location. For this purpose, data from the country wide value testing trials from 1991 to 2001 carried out by the Federal Office of Plant Varieties (Bundessortenamt) were used. Winter wheat cultivation regions were adopted according to the German convention which gives the chance of further dividing the data into regional subgroups. The results of the analyses indicate that for a given location, the two years regional data have the highest predictive power for superior cultivars. Two years’ data from that specific location give the highest predictive power for intermediate and inferior cultivars. In general, the predictive power of single year data is much lower than of two years data. The results confirm the merit of the definition of different cultivation regions. By proper definition of regions, the multiple year data collected within the region have high predictive power for the cultivar performance for the locations within that region.     

基于物候特征参数的橡胶树种植信息遥感提取研究?

为了更好地监测橡胶树的生长状态,客观准确地获取橡胶树的种植的地理位置信息,采用非对称高斯函数对海南岛内MODIS EVI时间序列数据进行滤波处理,定义作物物候特征参数并计算海南岛每个像元的特征参数值;统计分析基于MODIS EVI数据的海南岛橡胶树样区作物关键物候特征参数值变化规律;根据橡胶树样区特征参数值变化规律建立海南岛橡胶树种植信息提取算法,提取结果采用不同比例的橡胶树混合像元分类法表示。根据提取算法对海南岛2011年橡胶树种植信息进行提取,用验证数据对该提取算法进行精度检验与实用性评价,检验结果显示该算法的平均分类精度达85.97%,可以用来提取橡胶树种植信息。     

一种新的符合度算法及其应用

在总结分析了几种常用综合评价方法的基础上, 提出了一种反映观察值与理论值之间相似性的新算法--符合度。该算法就评价信息个体(观察值)与标准值(期望值)的马氏距离, 再由马氏距离转化为评价对象与标准的接近程度, 即符合度(r)。首先进行指标数(p)、相似度(r)与马氏距离(d)的模拟试验, 再通过曲面拟合的方法找出它们之间的关系模型。通过大量抽样试验, 验证符合度的次数分布与原先设定的符合度的良好对应关系, 说明模型的可行性与可靠性。以小麦RVA性状为指标, 利用该算法分析扬麦系统若干品种之间的接近程度, 并评价多变数复杂效应回归分析模拟试验的结果。符合度算法不需要数据标准化处理, 直接利用原始数据, 减少了计算工作量, 降低了因数据标准化处理方法不同而引起的评价结果差异, 同时由于不需要赋权, 排除了主观性的影响, 保证了信息的完整性以及评价结果的可靠性。