基于分水岭变换与空间聚类的高分辨率遥感影像面向对象分类

面向对象方法已广泛应用于高分辨率遥感影像分类,提出一种结合改进分水岭变换与空间聚类的遥感影像面向对象分类新方法。首先,基于相位一致思想分析图像特征,由Gabor小波多尺度、多方向提取QuickBird全色影像的梯度信息;利用扩展最小变换与强制最小技术分别获取图像前景标识、重建相位一致梯度图像,利用改进后的分水岭变换获得分割对象。然后,提取各对象的多波段光谱特征,利用Gabor小波获取对象纹理矢量,并用独立成分分析方法进行特征选择,依次进行对象的光谱与纹理聚类。最后,通过分析对象间空间拓扑关系判断聚类后不确定对象的类别属性。实验结果表明该方法能取得较好结果,在一定程度上提高了影像分类的自动化水平。     

A model for predicting resuspension of Escherichia coli from streambed sediments

To improve the modeling of water quality in watersheds, a model is developed to predict resuspension of Escherichia coli from sediment beds in streams. The resuspension rate is expressed as the product of the concentration of E. coli attached to sediment particles and an erosion rate adapted from work on sediment transport. The model uses parameter values mostly taken from previous work, and it accounts for properties of the flow through the bottom shear stress and properties of the sediment through the critical shear stresses for cohesive and non-cohesive sediment. Predictions were compared to resuspension rates inferred from a steady mass balance applied to measurements at sixteen locations in a watershed. The model’s predictions matched the inferred rates well, especially when the diameter of particles to which E. coli attach was allowed to depend on the bottom shear stress. The model’s sensitivity to the parameters depends on the contributions of particle packing and binding effects of clay to the critical shear stress. For the current data set, the uncertainty in the predictions is controlled by the concentration of E. coli attached to sediment particles and the slope used to estimate the bottom shear stress.     

集成U-Net方法的无人机影像胡杨树冠提取和计数

塔里木河流域的胡杨林是该荒漠区域典型的森林资源，胡杨树冠大小和株数信息对塔里木河流域森林资源监测、生态保护和恢复具有重要意义。由于该流域乔灌草植物群落分布的复杂性，传统方法很难实现胡杨树冠的精准分割和大范围的株数提取。以塔里木河中游胡杨林为研究区，选取几块典型胡杨林区域，提出集成深度学习和分水岭分割的处理方法，对密集胡杨树冠的精准分割和单株胡杨的提取进行了深入探讨。首先，将无人机影像（空间分辨率0.16 m）无缝拼接生成正射影像；采用U-Net卷积神经网络对胡杨树冠覆盖区域进行精准分割；在U-Net模型分割的基础上使用标记分水岭方法对密集胡杨树冠进行自动再分割和单株计数，计算出所选研究区的胡杨株数并精准定位。结果表明U-Net卷积神经网络对胡杨的所有树冠区域提取的平均精度可达94.1%，在胡杨树冠覆盖区域识别分割的基础上，用标记分水岭分割方法对胡杨单木计算总体精度为93.3%。研究认为，结合深度学习和标记分水岭方法为自动化大范围森林资源监测提供了新思路和借鉴经验。     

基于形态学梯度重建的分水岭算法改进研究

图像分割是图像分析和计算机视觉领域中一个最重要的处理过程,常用的边缘检测算子存在边缘定位不准等缺点,而传统的分水岭算法则对噪声敏感且存在过分割的问题。本研究提出基于形态学梯度重建的改进型分水岭算法,该算法通过在人脸图像预处理过程中对图像进行中值滤波消除部分噪声,然后再对滤波后的图像进行增强对比度以及重构梯度图处理,最后利用分水岭算法对处理后的图像进行分割。仿真实验应用不同的算法对同一图像进行计算分割,对比结果显示,本研究提出的方法能够较好地抑制过分割,并且能够分割出人脸图像的主要特征。     

Hydrogeographical Regionalisation: An Approach for Evaluating the Effects of Land Cover Change in Watersheds. A Case Study in the Cuitzeo Lake Watershed, Central Mexico

Efficient natural resources management, including continental water at watershed level, requires understanding the arrangement of landscape attributes in a region. The geographical analysis of landscape attributes is a useful approach to delineate relatively homogeneous watersheds or regions. This research was carried out in order to evaluate the effect of land cover through time on regionalisation modelling in the poorly-gauged Cuitzeo Lake Watershed, and to develop models to create two hydrogeographical regionalisations for the years 1975 and 2000. The inputs required by the regionalisation methodology were integrated in a GIS and validated before carrying out statistical regionalisation procedures (cluster analysis and PCA). GIS operations were done in Arc View 3.2 and statistical analyses in PC-ORD. Median Euclidean distances with mean distance linkage methods were used. A 75% of similarity was chosen as the threshold to generate regions. Importance rankings of regions were obtained using multicriteria evaluation methods. Based on the analysis, 38 of the 52 subwatersheds belonging to the Cuitzeo Lake Watershed were clustered; eight groups were defined in 1975 and nine in 2000. According to PCA, the strongest positive variables are associated to morphometric, geologic, land cover and soil attributes. Fourteen subwatersheds were never clustered. Four subwatersheds changed of cluster between 1975 and 2000. This change is explained because of rainfed agricultural parcel abandonment, and subsequent shrubland growth. This means that the hydrogeographical regionalisation is sensitive to land cover change processes. The methodological approach applied in this research is a low-cost and fast alternative for evaluating the impact of land cover and land use change on hydrogeographical regionalisation; in consequence, data and information generated during the analysis were made available to local authorities so that they can improve both water resources planning and their informational baseline for decision making and for development of environmental policies in the Cuitzeo Lake Watershed.     

降雨输入对青龙河流域HSPF模型模拟结果的影响

为评价降雨输入对青龙河流域BASINs/HSPF模型模拟结果的影响,改善HSPF模型模拟精度,应用趋势成分建模、周期成分建模、相依随机建模和独立随机建模（白噪声）等随机建模程序和蒙特卡洛计算机模拟方法,获得200组降雨随机模拟序列,分别作为青龙河流域BASINs/HSPF模型的输入,以Nash-Sutcliffe效率系数（E_NS）作为模型模拟效果的评价标准,获得如下结论:采用趋势成分、周期成分与ARMA建模以及正态随机模拟获得降雨随机模拟序列是量化降雨输入随机性的可行方法;在模型参数优化的条件下,降雨随机模拟序列HSPF模拟E_NS值的变化区间为[71.09%,74.96%],波动幅度达3.87%,表明降雨输入随机性对于HSPF模拟结果具有显著影响;当考虑每年的日降雨量极大值时,E_NS值变化区间为[75.35%,78.81%],波动幅度达3.46%,且结果均优于没有考虑降雨极大值点的降雨随机模拟序列,表明降雨时间序列的极大值点对于HSPF模拟效果具有显著影响;随着降雨时间序列中所考虑极大值点数量的逐渐增多,HSPF模拟效果出现下降趋势,表明HSPF模拟应特别关注若干最大极值点的影响;降雨输入的不确定性是HSPF模型模拟结果不确定性的重要来源,改善HSPF模拟效果需要考虑降雨时间序列随机性和极值点因素的影响.本研究可为量化降雨输入对HSPF模型模拟的影响以及HSPF模拟的降雨情景优选提供借鉴.     

Region-based Statistical Analysis of 2D PAGE Images

A new comprehensive procedure for statistical analysis of two-dimensional polyacrylamide gel electrophoresis (2D PAGE) images is proposed, including protein region quantification, normalization and statistical analysis. Protein regions are defined by the master watershed map that is obtained from the mean gel. By working with these protein regions, the approach bypasses the current bottleneck in the analysis of 2D PAGE images: it does not require spot matching. Background correction is implemented in each protein region by local segmentation. Two-dimensional locally weighted smoothing (LOESS) is proposed to remove any systematic bias after quantification of protein regions. Proteins are separated into mutually independent sets based on detected correlations, and a multivariate analysis is used on each set to detect the group effect. A strategy for multiple hypothesis testing based on this multivariate approach combined with the usual Benjamini-Hochberg FDR procedure is formulated and applied to the differential analysis of 2D PAGE images. Each step in the analytical protocol is demonstrated by using an actual dataset. The effectiveness of the proposed methodology is shown using simulated gels in comparison with the commercial software packages PDQuest and Dymension. We also introduce a new procedure for simulating gel images.     

一种CT医学图像分割新算法

针对临床辅助诊断的需要,提出了一种结合传统分水岭算法和新型核聚类算法的CT医学图像分割新算法。首先,通过分水岭变换,CT图被分割成不同的小区域。然后,根据改进的KFCM算法,利用Mercer核将各个小区域的平均灰度值映射到高维特征空间,使得原来在分水岭算法分割图中未显示出来的特征显现出来。通过此方法,相较于传统核聚类(KFCM)算法,我们可实现更准确的聚类,并有效解决分水岭算法分割CT医学图像的过分割问题,因此能取得更好的分割效果。实验结果显示,本文方法能够很好分割腹腔CT图,获得更清晰的分割图像。     

分水岭算法应用于主动轮廓模型能量分割算法的研究

主动轮廓模型(snake模型)被广泛应用于边缘提取、图像分割等领域。该模型能对目标适当初始化,并进行自主收敛,使得能量处于极小值状态,以达到目标分离的效果。当目标初始位置敏感时,需要依赖其他机制对内部能量进行合理初始化,由于模型的非凸性,它有可能收敛到局部极值点甚至发散。将分水岭算法应用于主动轮廓模型的能量分割算法,通过改进的分水岭算法确定主动轮廓模型的初始轮廓,利用迭代完成对轮廓点周围的局部近邻点的检索,以选取更小的轮廓模型,当获得最小值时完成目标轮廓的提取。     

生成式对抗网络下路面裂缝图像检测识别仿真

为了有效提高生成式网络下路面裂缝检测精度,针对路面裂缝图像的复杂多样性,提出一种基于分水岭算法的路面裂缝图像检测识别方法。通过对路面裂缝图像进行分析,采用邻近区域相关知识对路面裂缝图像进行虚假裂缝去除,提取出路面真实裂缝纹理图像,引用分水岭算法对裂缝图像进行分割,获取细化后的检测图像,并将相同路面的两种图像进行融合,完成路面裂缝图像检测识别。实验结果表明,所提出基于分水岭算法的路面裂缝图像检测识别方法检测时间较短、检测准确率较高、误报率较低。