基于人工神经网络的叶脉信息提取——植物活体机器识别研究Ⅰ

叶片的识别是识别植物的重要组成部分,特别在野外识别植物活体尤其重要。叶脉的脉序是植物的内在特征,包含有重要的遗传信息。但由于叶脉本身的多样性,利用单一特征的图像处理方法难以有效地提取叶脉。为了充分利用图像的信息,本文提出了一种基于人工神经网络的叶脉提取方法。该方法利用边缘梯度、局部对比度和邻域统计特征等10个参数来描述像素的邻域特征,并将其作为神经网络的输入层。实验结果表明,与传统方法相比,经过训练的神经网络能够更准确地提取叶脉图像,为进一步的叶片识别打下了良好的基础。     

植物叶形的计算机识别系统

植物叶形是识别植物的重要和常用形态特征,建立计算机自动识别系统对于认识和正确识别植物十分重要.本文论述了植物叶形图像识别系统设计中的图像处理、特征提取及分类识别等问题.本系统采用Visual Basic.Net编程工具设计,在Windows 2000/XP平台上通过叶片图像的输入、变换、平滑和分割等识别过程,实现了叶片图像的形状和叶缘特征的结果输出.实验结果表明,该系统能够很好地识别植物的叶形,对14种植物337份叶片样本的叶形测试准确率达93.2%.为植物识别的进一步研究奠定了基础.     

基于遥感的植物叶脉功能性状计算及其生态学意义

叶脉网络性状能够反映出植物适应特定生境的基本方式及其光合生理功能, 体现了植物水分和能量权衡的生态策略。叶脉功能性状的高效提取, 为植物生理功能和资源配置的响应机理研究提供参考。以北京市常见绿化树种洋白蜡(Fraxinus pennsylvanica)、臭椿(Ailanthus altissima)和国槐(Sophora japonica)的叶脉显微图像为训练样本, 采用遥感图像处理软件, 对叶脉显微图像进行多尺度分割和分类识别, 根据目标对象的亮度(brightness)、光谱(spectrum)和几何(shape)特征规则实现大批量叶脉的快速解译。结果表明, 3个树种叶脉分割的最优参数及自动提取规则为: 尺度参数(scale parameter)为200, 形状参数(shape parameter)为0.8, 紧凑度参数(compactness parameter)为0.2, 亮度值为190—230, 绿光波段大于210, 形状–密度指数为1.5。该方法提取叶脉密度(leaf vein density, LVD)和叶脉面积(leaf vein area, LVA)精度分别达到了95.7%和94.5%以上, 对该3种植物叶脉性状的快速提取具有较高的普适性。在城市不同环境中, 随着温度的升高、土壤水分含量的降低, 叶脉密度总体上呈增大的趋势, 而气孔数量则明显减少, 叶脉密度与气孔密度间普遍存在显著的负相关关系(P<0.05)。这说明了植物在高温和干旱胁迫环境中, 在叶片水平上表现出“此消彼长”的权衡规律, 进一步提高其逆境耐受性。     

基于遥感技术植物叶脉性状的计算及其生态学意义

叶脉网络性状能够反映出植物适应特定生境的基本方式及其光合生理功能,体现了植物水分和能量权衡的生态策略。叶脉功能性状的高效提取,为植物生理功能和资源配置的响应机理研究提供参考。以北京市常见绿化树种洋白蜡(Fraxinus pennsylvanica)、臭椿(Ailanthus altissima)和国槐(Sophora japonica)的叶脉显微图像为训练样本,采用遥感图像处理软件,对叶脉显微图像进行多尺度分割和分类识别,根据目标对象的亮度(brightness)、光谱(spectrum)和几何(shape)特征规则实现大批量叶脉的快速解译。结果表明,3个树种叶脉分割的最优参数及自动提取规则为:尺度参数(scale parameter)为200,形状参数(shape parameter)为0.8,紧凑度参数(compactnes sparameter)为0.2,亮度值为190—230,绿光波段大于210,形状–密度指数为1.5。该方法提取叶脉密度(leaf vein density,LVD)和叶脉面积(leaf vein area,LVA)精度分别达到了95.7%和94.5%以上,对该3种植物叶脉性状的快速提取具有较高的普适性。在城市不同环境中,随着温度的升高、土壤水分含量的降低,叶脉密度总体上呈增大的趋势,而气孔数量则明显减少,叶脉密度与气孔密度间普遍存在显著的负相关关系(P0.05)。这说明了植物在高温和干旱胁迫环境中,在叶片水平上表现出"此消彼长"的权衡规律,进一步提高其逆境耐受性。     

植物叶形的计算机识别系统

植物叶形是识别植物的重要和常用形态特征, 建立计算机自动识别系统对于认识和正确识别植物十分重要。本文论述了植物叶形图像识别系统设计中的图像处理、特征提取及分类识别等问题。本系统采用Visual Basic.Net编程工具设计, 在Windows 2000/XP平台上通过叶片图像的输入、变换、平滑和分割等识别过程,实现了叶片图像的形状和叶缘特征的结果输出。实验结果表明, 该系统能够很 好地识别植物的叶形, 对14种植物337份叶片样本的叶形测试准确率达93.2%。为植物识别的进一步研究奠定了基础。     

基于深度学习的葡萄砧木叶片识别研究

嫁接有利于增强树体对生物及非生物胁迫的适应能力,提高葡萄产量和品质。葡萄砧木品种多样复杂,识别难度较大,深度学习能够快速提取图像的深层特征,被广泛应用于植物图像分类识别领域。本研究以30份葡萄砧木成龄叶图像作为研究对象,通过采集叶片图像,构建了一个包含13547张的葡萄砧木叶片图像的数据集。采用GoogleNet、ResNet-50、ResNet-101以及VGG-16等4个卷积神经网络对其进行自动识别。结果表明：精度最高的分类网络为ResNet-101,在最优模型参数（学习率：0.005,最小批次：32,迭代次数：50）下精度达到97.5%。ResNet-101模型检测的30个品种中,平均预测精确率为92.59%,有7个品种的预测精确率达到100%;平均召回率为91.08%,有8个品种的召回率达到100%,叶片的叶面纹理、叶脉以及叶缘部分对品种识别的影响最大。以上结果证实,深度学习网络模型可以实现对葡萄砧木的自动实时识别,为葡萄砧木品种的保护、利用、分类研究以及其他农作物的品种识别提供参考。     

高寒植物叶片性状对模拟降水变化的响应

降水变化对高寒草甸生态系统产生了显著影响,植物叶片性状特别是叶脉特征对降水变化非常敏感,然而高寒植物叶片性状特征如何响应降水变化还知之较少。采用集雨棚模拟增减50%降水的条件,以高寒草甸8种主要植物叶片为研究对象,研究了降水变化对叶片的叶脉率、叶脉密度、叶片大小、比叶质量、叶片总有机碳含量、叶片全氮含量、叶片碳同位素相对含量和碳氮比等叶片性状的影响。发现增水显著增加了植物的叶片大小、稳定碳同位素千分值、总有机碳含量、全氮含量,但显著降低了叶脉密度;而减水显著降低了叶片大小、稳定碳同位素千分值。植物叶片性状各指标对降水变化的响应存在协同变化和相互制约。不同水分生态类型的植物对降水变化的响应存在差异,中生植物通过增加叶片大小和减少叶脉密度积极应对降水的增加,矮生嵩草的叶片大小分别增加了200.3%,叶脉密度减小了17.5%,而旱中生植物通过减少叶片大小和增加叶脉密度应对降水的减少,垂穗披碱草和异针茅的叶片大小分别减少54.9%和30.7%,其叶脉密度分别增加25%和22.4%。羽状叶脉植物增加叶脉密度和稳定碳同位素千分值以适应增水条件,花苜蓿、异叶米口袋的叶脉密度的增加了7.8%和4.0%,稳定碳同位素千分值增加2.5%和3.3%,但增水条件下平行叶脉植物的叶脉密度不变或降低和稳定碳同位素千分值保持不变;减水增加了平行叶脉植物叶脉密度并减低了稳定碳同位素千分值,异针茅的叶脉密度增加了22.4%,稳定碳同位素千分值减小2.9%,而对羽状叶脉植物的叶脉密度和稳定碳同位素千分值减少或不变。植物叶片性状对增水的敏感性显著大于对减水的敏感性,增水的效应约为减水的2倍;叶片大小的敏感性显著大于其它叶片性状,约为其它叶片性状的10倍。因此,植物在应对短期降水变化时,植物形态可塑性的作用凸显,放大或缩小叶片大小是植物应对降水变化的最有效的途径,但是不同水分生态类型和叶脉类型植物可塑性的方向存在显著差异。     

植物叶片末级小脉类型的识别

植物叶片末级小脉类型的识别张莉枝林月惠植物叶片包括皮系统、维管系统和基本组织系统。叶子的维管系统通称叶脉,并逐级分枝布满在整个叶片之中,叶脉不仅与基本组织系统中的叶肉细胞密切联系,而且还与叶柄中的维管系统相连接。叶脉的主要功能是进行水分和养分的双向运...     

叶脉网络功能性状及其生态学意义

叶脉网络结构是叶脉系统在叶片里的分布和排列样式。早期叶脉网络结构研究主要集中在其分类学意义上; 近年来叶脉网络功能性状及其在植物水分利用上的意义已成为植物生态学研究的热点。该文介绍了叶脉网络功能性状的指标体系(包括叶脉密度、叶脉直径、叶脉之间的距离、叶脉闭合度等), 综述了叶脉网络功能性状与叶脉系统功能(包括水分、养分和光合产物等物质运输、机械支撑和虫害防御等)的关系, 叶脉网络功能性状与叶片其他功能性状(包括比叶重、叶寿命、光合速率、叶片大小、气孔密度等)的协同变异和权衡关系, 以及叶脉网络功能性状随环境因子(包括水分、温度、光照等)的变化规律等方面的最新研究进展。此外, 叶脉网络功能性状的研究成果也被应用于古环境重建、城市交通规划、流域规划及全球变化研究中。由于叶脉网络功能性状是环境因子与系统发育共同作用的结果, 未来开展分子—叶片—植物—生态系统等多尺度的叶脉网络功能性状研究, 理清叶脉网络功能性状与气孔失水—茎干导水—根系吸水等植物水分利用的关系, 将为预测植物及生态系统对全球变化的响应提供新的启示。     

植物叶脉标本简单制作方法及基于Image J软件的叶脉图像分析

采用5%氢氧化钾沸水浴对12种植物叶片进行处理,结合超声波清洗使叶肉及表皮脱离,经漂白、染色制成叶脉标本。使用数码相机及显微镜对标本进行成像,利用Image J软件MINA工具宏的分析步骤对图像进行分析,测定12种植物叶脉的密度、分支密度、交点密度、端点密度,并通过手动追踪验证测量结果的准确性。结果表明,该方法可快速获得清晰完整的叶脉网络结构图,并且可以运用Image J软件MINA工具宏的分析步骤对叶脉进行分析。与传统的氢氧化钠组织透明法比较,该方法处理时间短,叶脉网络结构完整清晰,用Image J软件可准确测量叶脉密度。     

Evolution and Function of Leaf Venation Architecture: A Review

The leaves of extant terrestrial plants show highly diverseand elaborate patterns of leaf venation. One fundamental featureof many leaf venation patterns, especially in the case of angiospermleaves, is the presence of anastomoses. Anastomosing veins distinguisha network topologically from a simple dendritic (tree-like)pattern which represents the primitive venation architecture.The high degree of interspecific variation of entire venationpatterns as well as phenotypic plasticity of some venation properties,such as venation density, indicate the high selective pressureacting on this branching system. Few investigations deal withfunctional properties of the leaf venation system. The interrelationshipsbetween topological or geometric properties of the various leafvenation patterns and functional aspects are far from beingwell understood. In this review we summarize current knowledgeof interrelationships between the form and function of leafvenation and the evolution of leaf venation patterns. Sincethe functional aspects of architectural features of differentleaf venation patterns are considered, the review also refersto the topic of individual and intraspecific variation. Onebasic function of leaf venation is represented by its contributionto the mechanical behaviour of a leaf. Venation geometry anddensity influences mechanical stability and may affect, forexample, susceptibility to herbivory. Transport of water andcarbohydrates is the other basic function of this system andthe transport properties are also influenced by the venationarchitecture. These various functional aspects can be interpretedin an ecophysiological context. Copyright 2001 Annals of BotanyCompany Review, leaves, leaf venation, evolution, network, transport, flow, mechanical stabilization     

CLASSIFICATION OF THE ARCHITECTURE OF DICOTYLEDONOUS LEAVES

A classification of the architectural features of dicot leaves—i.e., the placement and form of those elements constituting the outward expression of leaf structure, including shape, marginal configuration, venation, and gland position—has been developed as the result of an extensive survey of both living and fossil leaves. This system partially incorporates modifications of two earlier classifications: that of Turrill for leaf shape and that of Von Ettingshausen for venation pattern. After categorization of such features as shape of the whole leaf and of the apex and base, leaves are separated into a number of classes depending on the course of their principal venation. Identification of order of venation, which is fundamental to the application of the classification, is determined by size of a vein at its point of origin and to a lesser extent by its behavior in relation to that of other orders. The classification concludes by describing features of the areoles, i.e., the smallest areas of leaf tissue surrounded by veins which form a contiguous field over most of the leaf. Because most taxa of dicots possess consistent patterns of leaf architecture, this rigorous method of describing the features of leaves is of immediate usefulness in both modern and fossil taxonomic studies. In addition, as a result of this method, it is anticipated that leaves will play an increasingly important part in phylogenetic and ecological studies.     

Classification of Camellia (Theaceae) species using leaf architecture variations and pattern recognition techniques

Leaf characters have been successfully utilized to classify Camellia (Theaceae) species; however, leaf characters combined with supervised pattern recognition techniques have not been previously explored. We present results of using leaf morphological and venation characters of 93 species from five sections of genus Camellia to assess the effectiveness of several supervised pattern recognition techniques for classifications and compare their accuracy. Clustering approach, Learning Vector Quantization neural network (LVQ-ANN), Dynamic Architecture for Artificial Neural Networks (DAN2), and C-support vector machines (SVM) are used to discriminate 93 species from five sections of genus Camellia (11 in sect. Furfuracea, 16 in sect. Paracamellia, 12 in sect. Tuberculata, 34 in sect. Camellia, and 20 in sect. Theopsis). DAN2 and SVM show excellent classification results for genus Camellia with DAN2's accuracy of 97.92% and 91.11% for training and testing data sets respectively. The RBF-SVM results of 97.92% and 97.78% for training and testing offer the best classification accuracy. A hierarchical dendrogram based on leaf architecture data has confirmed the morphological classification of the five sections as previously proposed. The overall results suggest that leaf architecture-based data analysis using supervised pattern recognition techniques, especially DAN2 and SVM discrimination methods, is excellent for identification of Camellia species.     

Venation networks and the origin of the leaf economics spectrum

The leaf economics spectrum describes biome-invariant scaling functions for leaf functional traits that relate to global primary productivity and nutrient cycling. Here, we develop a comprehensive framework for the origin of this leaf economics spectrum based on venation-mediated economic strategies. We define a standardized set of traits - density, distance and loopiness - that provides a common language for the study of venation. We develop a novel quantitative model that uses these venation traits to model leaf-level physiology, and show that selection to optimize the venation network predicts the mean global trait-trait scaling relationships across 2548 species. Furthermore, using empirical venation data for 25 plant species, we test our model by predicting four key leaf functional traits related to leaf economics: net carbon assimilation rate, life span, leaf mass per area ratio and nitrogen content. Together, these results indicate that selection on venation geometry is a fundamental basis for understanding the diversity of leaf form and function, and the carbon balance of leaves. The model and associated predictions have broad implications for integrating venation network geometry with pattern and process in ecophysiology, ecology and palaeobotany.     

Comparative anatomy of the young stem,node, and leaf of Bonnetiaceae,including observations on a foliar endodermis

A comprehensive study of the nodal and leaf anatomy of Bonnetiaceae was completed in order to provide evidence for evaluation in relation to systematics. Nodal anatomy is trilacunar, three-trace or unilacunar, one-trace. Basic leaf anatomical features of the family include: complete or incomplete medullated vascular cylinder in petiole; paracytic mature stomata with encircling ridges; large mucilaginous cells in the adaxial surface of mesophyll; periclinal divisions in upper surface layers; and discrete patches of phloem within the vascular bundles. Especially noteworthy is the presence in some genera of foliar vascular bundles enveloped by a sheath composed of two concentric regions, i.e., an inner region consisting of multiple layers of fibers and an outer specialized endodermis composed of thin-walled cells with Casparian strips. Leaves are variable with respect to lamina and cuticle thickness, relative amount and number of palisade and spongy layers, venation of lamina, and the presence or absence of sclereids and crystals in the mesophyll. A major feature in the evolution of Bonnetiaceae is development of a highly divergent, essentially parallel, leaf venation that is superficially similar to that of some monocotyledons and apparently unique among dicotyledons. Foliar anatomy provides important characters for the recognition of subgroups within Bonnetiaceae and is consistent with the segregation of Bonnetiaceae from Theaceae.     

Topological Phenotypes Constitute a New Dimension in the Phenotypic Space of Leaf Venation Networks

The leaves of angiosperms contain highly complex venation networks consisting of recursively nested, hierarchically organized loops. We describe a new phenotypic trait of reticulate vascular networks based on the topology of the nested loops. This phenotypic trait encodes information orthogonal to widely used geometric phenotypic traits, and thus constitutes a new dimension in the leaf venation phenotypic space. We apply our metric to a database of 186 leaves and leaflets representing 137 species, predominantly from the Burseraceae family, revealing diverse topological network traits even within this single family. We show that topological information significantly improves identification of leaves from fragments by calculating a “leaf venation fingerprint” from topology and geometry. Further, we present a phenomenological model suggesting that the topological traits can be explained by noise effects unique to specimen during development of each leaf which leave their imprint on the final network. This work opens the path to new quantitative identification techniques for leaves which go beyond simple geometric traits such as vein density and is directly applicable to other planar or sub-planar networks such as blood vessels in the brain.     

中国连香树科的系统研究——Ⅰ.叶的宏观结构及叶柄维管束变化

本文研究了连香树科(Cercidiphyllaceae)叶的宏观结构,首次报道连香树齿腺体显微结构及晶体类型,并对叶柄维管束的变化作了进一步研究。通过与近缘科的比较,我们认为连香树科的系统演化处于孤立地位,和金缕梅科有较近的亲缘关系,与木兰科较为疏远。     

Leaf extraction and analysis framework graphical user interface: segmenting and analyzing the structure of leaf veins and areoles

Interest in the structure and function of physical biological networks has spurred the development of a number of theoretical models that predict optimal network structures across a broad array of taxonomic groups, from mammals to plants. In many cases, direct tests of predicted network structure are impossible given the lack of suitable empirical methods to quantify physical network geometry with sufficient scope and resolution. There is a long history of empirical methods to quantify the network structure of plants, from roots, to xylem networks in shoots and within leaves. However, with few exceptions, current methods emphasize the analysis of portions of, rather than entire networks. Here, we introduce the Leaf Extraction and Analysis Framework Graphical User Interface (LEAF GUI), a user-assisted software tool that facilitates improved empirical understanding of leaf network structure. LEAF GUI takes images of leaves where veins have been enhanced relative to the background, and following a series of interactive thresholding and cleaning steps, returns a suite of statistics and information on the structure of leaf venation networks and areoles. Metrics include the dimensions, position, and connectivity of all network veins, and the dimensions, shape, and position of the areoles they surround. Available for free download, the LEAF GUI software promises to facilitate improved understanding of the adaptive and ecological significance of leaf vein network structure.     

Approaches to the identification of angiosperm leaf remains

During the past 125 years the history of early angiosperms, interpreted through the fossil leaf record has been largely an exercise in paleofloristic studies, ignoring evolution. Imprecise identifications of ancient leaves “matched” to extant genera and families have been used as the basis for reconstructions of paleocommunities and paleoclimates. However, as the result of careful morphological studies of leaf form, venation and cuticular features new insights into the evolution of angiosperms are now available. In this paper considerations are given to the usefulness and shortcomings of leaf form, venation and cuticular analysis as diagnostic tools of plant identification. Many techniques for the study of the morphology of modern and fossil leaves are included in this paper as well as tables outlining features of leaf venation and the epidermis. Careful morphological studies of leaf form (such as the venation and epidermal characters emphasized in this paper) will provide better understanding of the relationships of living angiosperms and transform the fossil leaf record into useful data that can be used to study the evolution of the angiosperms.