复杂背景下黄瓜病害叶片的分割方法研究

利用图像处理和模式识别技术进行复杂背景下黄瓜叶部病害的自动识别,需要先把目标叶片从复杂背景中分割出来,才能进行后续的特征提取和病害识别。为实现复杂背景下黄瓜叶片的分割,首先利用K-均值聚类算法去除图片中的非绿色部分,再采用基于laplacian of gaussia(LOG)算子的方法对待分割的叶片进行区域检测,然后进行基于形状上下文(shape context)的模板匹配和分割。为了提高匹配速度,先检测叶片的生长点和叶尖,以确定叶片的位置、尺寸和方向;然后使用基于超像素(superpixel)的最优匹配搜索方法来减少搜索的复杂度。对20幅黄瓜叶部病害图像进行分割测试,并与人工分割法进行对比,结果表明,本文所采用的分割算法能较好地从复杂背景下提取出黄瓜叶部病害图像,分割准确率达94.7%,为后期黄瓜病斑的特征提取等工作奠定了良好的基础。     

AFLP-based genetic linkage map of marine shrimp Penaeus (Fenneropenaeus) chinensis

This paper presents the genetic linkage map of the Chinese shrimp Penaeus (Fenneropenaeus) chinensis constructed with 472 AFLP markers. A hundred F₁ progeny from an intercross between a female from the new variety “Yellow Sea No. 1” and wild caught male used for the mapping study. Two separate maps were constructed for each parent. The female linkage map consisted of 197 marker loci forming 35 linkage groups and spanned a total length of 2191.1 cM, with an average marker space of 13.5 cM. The male map consisted of 194 marker loci mapped to 36 linkage groups and spanned a total length of 1737.3 cM, with an average marker spacing of 11.0 cM. The level of segregation distortion observed in this study was 12.2%. The estimated genome length of P. chinensis was 3150.3 cM for the female and 2549.3 cM for the male, respectively. The observed genome coverage was 69.6% for the female and 68.1% for the male map. The linkage maps constructed in this study provide basic information for further linkage studies on Chinese shrimp, and more importantly, the construction of the maps are part of the work of the genetic breeding programs which will be used for growth discovered in the QTL analysis of P. chinensis.     

Interactive mapping of urban tree benefits using Google Fusion Tables and API technologies

Geographic Information Systems (GIS) and web-based GIS platforms are typically used by large organizations and government agencies for the purpose of data storage, query, analysis, and spatial data mapping. Google Fusion Tables, however, provides cloud-based computing services for data management and easy user collaboration through the Google Maps interface. Fusion Tables are oriented towards smaller organizations that previously were unable to publish data online due to limitations of database knowledge, high cost of start-up, and restrictions on implementation. In this research project, we assess the functionality and limitations of Fusion Tables and associated API as a spatial data management system by programming an interactive web-based mapping platform, the CityTrees.ca Project, designed to showcase the benefits of trees growing on the campus of Ryerson University, Toronto. Specifically, Fusion Tables functionality, such as data storage and manipulation, is used to build a query interface (both attribute and spatial) that permits users of CityTrees.ca to search for and map trees by species, diameter, height, and location. We found that Fusion Tables performed well as a storage medium for our campus tree data, which could easily be explored through our creation of a JavaScript-enabled query tool. Our project provides a unique roadmap for small to medium-sized urban forestry organizations and environmental non-profits seeking to create interactive mapping applications where data privacy and advanced spatial analysis are not fundamental.     

利用重组自交系群体定位玉米生育期相关性状QTL

利用玉米自交系80007和80044为亲本,衍生包含355个家系的重组自交系(RIL)F9群体,采用SSR标记构建包括219个标记的连锁图谱,图谱总长度2 133.5 cM,标记间平均距离9.7 cM。采用完备区间作图法对控制玉米抽雄期、散粉期、吐丝期以及散粉至吐丝间隔期的4个生育期相关性状进行QTL分析。结果表明,共检测到60个QTL,其中抽雄期检测到20个QTL,吐丝期检测到15个QTL,散粉期检测到20个QTL,散粉至吐丝间隔期检测到5个QTL。共有7个QTL对表型变异的解释率超过了10%,表现为主效QTL效应,分布在第3、4和第9染色体。有11个QTL在不同环境中能重复检测出,是受环境影响较小、为较稳定的QTL。分析发现,生育期相关性状的QTL在染色体上有"成簇"分布的现象,并且贡献率较大的QTL控制着多个相关性状。结果表明,bin3.04-3.05、bin3.09、bin7.03和bin9.02-9.03是生育期相关性状QTL的密集区域,这些区域存在对生育期相关性状重要作用的位点。     

采用改进Mask R-CNN算法定位鲜食葡萄疏花夹持点

为实现鲜食葡萄疏花机械化与自动化,该研究提出了一种鲜食葡萄疏花夹持点定位方法。首先基于ResNeXt骨干网络并融合路径增强,改进Mask R-CNN模型,解决鲜食葡萄花穗、果梗目标较小难以检测的问题;进而针对花穗、果梗生长姿态的复杂性与不确定性,提出一种集合逻辑算法,该算法采用IoU函数剔除重复检测的花穗与果梗,建立花穗、果梗对,并对果梗掩模进行形态学开运算,利用集合关系获取主果梗掩模,确定以主果梗质心附近的中心点为果梗的夹持点。最后,随机选取测试集中的图像进行试验。试验结果表明：果梗夹持点平均定位准确率为83.3%,平均定位时间为0.325 s,夹持点x、y方向定位误差及定位总误差最大值分别为10、12和16像素,能够满足鲜食葡萄疏花的定位精度与速度要求,可为实现鲜食葡萄疏花机械化与自动化提供理论支撑。     

基于改进边缘分割算法的幼苗信息提取

优化特征空间和改进分割算法是利用面向对象技术准确获取幼苗信息的关键,也是高空间分辨率数据提取目标地物信息迫切需要解决的问题。研究了在多光谱影像进行去噪声处理基础上,采用改进的基于边缘的算法进行影像分割,同时选取纹理、形状、光谱特征构建特征空间,实现幼苗信息提取的方法。结果表明,该方法对幼苗信息提取的总精度达86%,比传统技术提高了12%,KAPPA系数达0.814 5,比传统技术提高了0.115 9。该方法可以对幼苗信息进行准确快速提取,能够为生产或管理部门进行准确监测和决策提供依据,对未来造林情况进行预测和评价有重要意义。     

吉林省本地鹅和杂交鹅的产肉性能及胴体分割的研究

:随机选择刚出生的本地鹅雏和杂交鹅雏各 130只 ,采用放牧加补饲的方式育肥至 6 3日龄 ,进行屠宰测定。结果本地鹅平均活重 32 87 50g ,胴体重 2 854 10g ,半净膛率 80 18% ,全净膛率 73 76 % ,头重 12 9 3g ,颈重 2 4 3 2g ,翅重 157 6g ;杂交鹅平均活重 3482 6 9g ,胴体重 2 96 1 92g ,半净膛率 78 6 3% ,全净膛率 72 6 % ,头重 136 2g ,颈重 2 57 5g ,翅重 16 6 2g。表明该方式育肥效果显著 ,且杂交鹅优于本地鹅。通过与全放牧方式育肥的同龄本地鹅对比 ,说明品种、性别、育肥方式对育肥效果均有影响     

水蜡蛾中国属及其地理分布

本文将水蜡蛾科（Brahmaeidae）的中国种归属问题进行了统一，并把外生殖器特征纳入属征，系统描述了各届特征和在中国的分布。     

禽腺病毒贵州分离株ＨＳ—１的核酸酶切分析与蛋白抗原鉴定

本研究利用１０％ＳＤＳ－ＰＡＧＥ及Ｗestern-Blotting技术鉴定３株不同的减蛋综合征病毒（贵州株ＨＳ－１、南京分离毒ＧＣ２、国际标准毒ＡＶ－１２７）的蛋白抗原性，结果表明，３株毒的蛋白多肽分子量均在１０６－１２kＤ范围,一各条多肽的组成上略有差异；它们都具有两条相同的免疫原性多肽，分子量分别为１０６和１００ＫＤ。同时，运用ＥcoRⅠ,HindⅢ，ＰｓｔⅠ和ＳmaⅠ等５种限制性内切酶进行了酶切分析，证实３株ＥＤＳ病毒用ＰstⅠ和ＳmsⅠ酶切的片段大小及长度略有不同，而其它３种酶的酶切图谱完全相同。     

不同品系西方蜜蜂基因组DNA多态性图谱构建及杂交分析

本研究应用随机引物K（5'-CGGCCCCTGT-3'）对不同品系西蜂的基因组DNA扩增，获得了能区分不同品系西蜂的多态性DNA图谱，从多态性图谱中选取一个共同片段K1680bp，标记成探针，探针K1680bp与不同品系西蜂的基因组DNA匀有杂交，从而证明片段K1680bp是西蜂蜂种的共性，为西蜂的鉴定提供了直接依据。