计算机视觉技术在苹果叶片营养诊断上的应用

植物营养诊断方法很多,但是无论何种方法,由于复杂的检测技术条件,在实际应用中均存在着一定困难,人们希望获得更快更简捷的检测方法和工具.随着计算机视觉技术的专业化与计算机硬件成本的降低和速度的提高,其在农产品诊断领域应用计算机视觉已变得越来越具有吸引力.苹果叶片营养状态诊断是进行合理施肥的前提,因此具有非常重要的意义.     

基于图像处理的温室黄瓜霜霉病诊断系统

为进一步提高温室黄瓜霜霉病诊断的准确率,构建了一个基于图像处理的温室黄瓜霜霉病诊断系统。针对温室黄瓜栽培现场采集的病害图像,采用基于条件随机场(Conditional random fields,CRF)的图像分割方法进行病斑图像分割,并采用决策树模型扩展一元势函数,提高病斑图像分割的准确性;将分割后的病斑图像转换到HSV颜色空间并提取其颜色、纹理和形状等25个特征,利用粗糙集方法进行特征选择与优化;构建了基于径向基核函数的SVM分类器,准确地识别与诊断温室黄瓜霜霉病。系统试验验证结果表明,该系统采用的病斑分割方法,能够克服复杂背景和光照条件的影响,准确地提取病斑图像;采用粗糙集方法能够有效地选择分类特征,将25个初始特征减少到12个,提高了运行效率;黄瓜霜霉病识别准确率达到90%,能够满足设施蔬菜叶部病害诊断的需求。     

基于茎秆直径微变化的拔节期玉米水分亏缺诊断指标研究

为了探索利用茎秆直径微变化诊断玉米水分亏缺状况的最优指标,采用桶栽的方法,开展了不同水分处理条件下拔节期夏玉米茎秆直径微变化中的日最大收缩量(MDS)、日增长量(DI)和当日恢复时间(RT)3个指标的变化规律及其与土壤相对含水率之间相关关系方面的试验研究。结果表明,(1)不同处理间的MDS和DI受水分亏缺影响差异较为明显,RT对土壤相对含水率的响应较为一致;(2)玉米日最大收缩量MDS和日增长量DI受到环境因素的影响较大,与土壤相对含水率θ的相关系数较小;(3)恢复时间RT与土壤相对含水率θ的相关系数最高,且受水分亏缺处理的影响较小,可以作为诊断玉米水分亏缺状况的较好指标。综上可知,与MDS和DI相比,RT更适合诊断夏玉米水分亏缺状况。     

基于改进级联神经网络的大豆叶部病害诊断模型

针对大豆叶部病害性状特征与病种之间的模糊性和不确定性,将数字图像处理技术与神经网络智能推理技术相结合,充分挖掘大豆受病害胁迫后表现性状与病种之间的潜在规律,提出了基于改进级联神经网络的大豆病害诊断模型。首先利用自制载物模板无损采集大田大豆叶部病害数字图像,计算病斑区域的形状特征、颜色特征及纹理特征14维度特征参数;为突显各方面特征对于不同病害种类决定作用的差异性,构建各子神经网络并联的第1级网络,第2级网络的输入为第1级网络的输出,利用多维特征各自优势来自动取得病种模式推理规则,建立了用于大豆叶部病害自动诊断的两级级联神经网络模型,仿真实验准确率为97.67%;同时应用量子遗传计算优化级联神经网络参数,平均迭代次数为743,平均网络误差为0.000 995 445,提高了学习效率,实现了大豆叶部病害的高效自动诊断和精确测报,为大田农作物全面系统地开展作物病害监测、智能施药及自动防治提供了理论依据。     

基于Lucid多途径检索的油茶害虫诊断APP开发与实现

油茶是亚热带地区重要的木本油料作物,近年来随着种植面积的不断扩大和集约化经营,病虫害问题日益严重,对油茶的产量和品质造成不同程度的影响。本研究通过野外调查、文献查阅、种类鉴定等工作,整理了21种油茶主要害虫的形态特征、生物学习性、为害特点、防治措施等信息,并拍摄相应害虫及为害状的高清照片1 600张。采用Lucid智能诊断系统,构建Fact Sheet Fusion基础信息数据库,并针对多途径检索方式提取害虫"为害方式"、"为害部位"、"为害高峰期"、"形态特征"等4个1级特征组,以及10个2级特征和18个3级特征,共组成102个特征状态,构建了油茶害虫的快速诊断系统。在此基础上,转换得到基于Android系统的手机应用APP,为林农和森防一线工作者对油茶害虫快速识别和防治提供便捷服务,从而促进油茶产业持续健康发展。     

Seasonal distribution of phytoplasmas in Australian grapevines

The distribution and persistence of phytoplasmas were determined in Australian grapevines. Phytoplasmas could be detected using the polymerase chain reaction (PCR) from shoots, cordons, trunks and roots throughout the year, and phytoplasmas appear to persistently infect Australian grapevines from year to year. Phytoplasmas were not always detected in samples from the same sampling area from one sampling period to the next. Phytoplasma detection by PCR was improved by sampling from shoots, cordons and trunks, especially during October (early spring). The diseases expressed by the 20 grapevines used in the distribution and persistence studies were monitored. Australian grapevine yellows disease (AGY) was expressed by 17/20 grapevines at some time during the study, whilst only 4/20 and 15/20 grapevines expressed restricted growth disease (RG) and late season leaf curl disease (LSLC), respectively. All grapevines with RG and LSLC also had AGY. The three diseases were persistently expressed in some grapevines and remission of disease was observed in others. The results of PCR detection in the same grapevines indicated that phytoplasmas were more frequently detected in AGY-affected grapevines that also expressed RG and LSLC compared with grapevines expressing AGY alone. Phytoplasmas were detected in symptomless plant material but less frequently compared with AGY-affected material.     

Occurrence, Distribution and Epidemiology of Grapevine Yellows in Spain

The main viticultural production areas in Spain were surveyed in 1994, 1995 and 1996 for the occurrence and incidence of Grapevine Yellows diseases associated to phytoplasmas. Samples from 300 plants showing symptoms of phytoplasma infection were collected from grapevine fields in the Spanish regions of Aragón, Catalonia and Navarra and analysed by PCR with specific primers for a non-ribosomal DNA of stolbur/Bois Noir (BN) and of Flavescence dorée (FD) phytoplasma. Nested PCR with universal primers P1/P7 and fU5/rU3 was also used. In the survey conducted in 1994 and 1995 only BN/stolbur phytoplasma was detected. The incidence of symptomatic plants was low in five plots of Catalonia from 3% to 18% in 1994 and 1995, respectively, and high in two plots of Navarra, from 60% to 80%. In the survey conducted in 1996 the incidence of symptomatic plants in Catalonia increased (6–80%) due to the presence of FD in five plots in the Northeastern Catalonia. An epidemiological study was carried out in two BN-affected plots of two regions from 1994 to 1997, with the evaluation of potential vectors and of host plants. The stolbur phytoplasma was found in individuals from different insect species belonging to the families Cicadellidae and Delphacidae. Some wild plants naturally infected with stolbur phytoplasma around the infected grapevines were: Convolvulus arvensis, Lavandula officinalis, Polygonum convolvulus, Solanum nigrum, and Thymus officinalis. The incidence of the disease in one BN-infected grapevine plot increased from 3.4% in 1994 to 18.40% in 1997.     

定量分解气候变化与人类活动对季节径流变异的贡献率

变化环境下径流序列可能发生变异。当前径流变异的贡献率分解研究多集中于年尺度,而对季节尺度的关注不足。以黄土高原典型流域——泾河流域为研究对象,应用小波分析方法揭示其季尺度径流序列的演变规律,采用有序聚类法和累积距平法对其径流序列进行变异诊断,最后采用累计斜率变化率比较法定量分解气候变化和人类活动对其四季径流变异的贡献率。结果表明:(1)泾河流域夏季和冬季径流的周期变化均在20世纪90年代左右有发生转变,均由小尺度周期变化转变为较大尺度周期变化;(2)春、夏、秋和冬季径流分别在1969年、1996年、1985年和1990年发生变异;(3)人类活动对春、夏、秋和冬季径流变异的贡献率分别为50.40%,89.69%,39.14%和84.59%,在夏季和冬季人类活动(主要是水土保持措施的不断深入)对泾河径流变异的贡献占主导地位,且干扰影响了径流的周期变化。     

解决夏季母猪“饥而不饿”的营养招数

夏季来临,温度是影响猪群健康的第一要素,尤其对于母猪,在高温下其对营养的吸收利用及代谢均会发生改变,导致母猪代谢不畅而不愿采食,这将严重影响猪群健康及繁殖成绩。文章将从营养调节角度分析,并结合生产实践来解决这一问题,以期为猪场夏季母猪健康养殖提供借鉴和参考。     

2016/2017冬中国气温偏暖的基本特征及可能成因

利用中国740余站观测的月平均2 m气温资料以及NCEP/NCAR再分析资料,通过动力学诊断的方法,对2016—2017年中国冬季异常偏暖现象进行了研究。结果发现：中国2016—2017年冬季气温偏暖主要是由于西伯利亚高压和阿留申低压偏弱导致东亚冬季风偏弱,在中国存在异常的偏南风。阿留申低压偏弱由两个因素引起,一是由于阿留申群岛地区的海温偏低,冷源对应异常高压导致阿留申低压偏弱;二是阿留申低压在东亚副热带高空西风急流出口区以北,然而该冬季高空西风急流出口区位置偏南,导致阿留申地区偏离急流出口减压区,从而导致阿留申低压偏弱。此外,西风带偏北,中高纬亚洲大陆上空以纬向环流为主,不利于冷空气南下。