Skip navigation
Skip navigation

On growth and form of leaves in 3 dimensions: Applications of machine vision and advanced optics

Singh, Amit K.

Description

Leaves have long been considered one of the hallmarks of angiosperm evolution and diversification. Development of morphologically complex surfaces with diverse three-- shape is presumably optimized to balance conflicting functional demands. Driven by selection pressures, highly variable phenotypes expressed in different environments can be a result of genetic differences or phenotypic plasticity. In recent years, the role of biomechanical...[Show more]

dc.contributor.authorSingh, Amit K.
dc.date.accessioned2019-06-25T23:46:47Z
dc.identifier.otherb59286167
dc.identifier.urihttp://hdl.handle.net/1885/164225
dc.description.abstractLeaves have long been considered one of the hallmarks of angiosperm evolution and diversification. Development of morphologically complex surfaces with diverse three-- shape is presumably optimized to balance conflicting functional demands. Driven by selection pressures, highly variable phenotypes expressed in different environments can be a result of genetic differences or phenotypic plasticity. In recent years, the role of biomechanical factors such as regulated regional growth rate and the interaction between the leaf lamina and the venation network during growth have attracted significant attention. Decades of work illustrate a complex interaction of genetic and environmental factors that influence generation of leaf diversity. Recent investigations explore the role of mechanical, abiotic factors and the cost of producing and maintaining an efficient functioning leaf in evolution of leaf form. To investigate the true nature of leaf diversity and physiological relevance of three-dimensional leaf shape, noninvasive, quantitative methods are becoming increasingly relevant and state-ofthe art. First, I developed a novel photogrammetric imaging system and a simplified geometric quantification framework that rigorously analyses leaf shape in threedimensional space. Our technique decomposes the leaf surface in morphologically distinct regions, the three-dimensional configuration of leaf margins and the rest of the lamina. Our novel non-invasive, photograph based imaging system generates a point cloud of the leaf surface. By adapting low cost techniques such as structure from motion (SfM) and Multi view stereo (MVS), we track the growth of Arabidopsis thaliana leaves from early stages of growth to maturity. A sequence of overlapping images are processed and aligned to obtain a high-resolution 3D structure of the leaf in the form of a raw point cloud, which can be further processed to obtain a triangulated mesh (smooth surface). Together these two techniques of computer vision and geometric quantification allow us to generate accurate 3D leaf models of shape change in Arabidopsis thaliana leaves during growth. The form of the leaf margin of the growing leaf was quantified in terms of lobiness (in-plane curvature) and waviness (out of plane curvature) and the leaf lamina was quantified using global surface curvature features of Gaussian and mean curvature of the surface. A significant fluctuation in the curvatures is seen as leaf size increases; the lobiness and mean curvature of the leaves show marked changes in particular. Our results allow for systematic and accurate quantitative phenotyping of growing leaf shapes in three-dimensional space. v Second, inspired by astonishing diversity of leaves among species of the genus Pelargonium, I adapted our three-dimensional imaging system and geometric quantification for in-field ecophysiological applications. In outdoor conditions the system allows the generation of detailed leaf shape models with much lower effort than other techniques. In a unique in-field study we quantify threedimensional morphological features of 25 Pelargonium species by generating accurate leaf surface meshes. We apply landmark free leaf surface and marginal curvature descriptors to quantify leaf shape and generate a curvature based morphological lineage of genus pelargonium. This study shows that in Pelargonium leaves, leaf surface area and 3 dimensional shape of leaf lamina is tightly correlated but that there is a weak correlation with leaf lamina thickness. This indicates a prominent structural role of the mid-vein and the venation network. We show the capability of such a system to be combined with traditional ecophysiological measurements, which can work as a framework for investigating leaf form and its functional relevance. Next, to investigate the role of Mid-vein in leaf shape determination, I introduced a novel volumetric, live optical imaging platform to quantify mid-vein and lamina volume. To date, there are no available methods to quantify whole leaf volumetric growth of the mid-vein and the lamina, which could allow development of a growth model, which explores the role that the mid-vein may play as a structural determinant of shape. Most modern techniques for studying internal leaf structures are either logistically constrained to a small sample size or destroy the tissue sample. I developed an interferometry based imaging platform that bridges the resolution and live imaging gap and allows us to generate distinct volumetric measures of Arabidopsis thaliana mid-vein and lamina. Our phenotyping platform obtains a deep tissue, high-resolution imaging with penetration depth of ~ 600 microns in live leaf samples. Our results demonstrate that such a system can be incorporated as an across-scale phenotyping method allowing fast, high-resolution volumetric quantification of leaf growth, structural and morphological traits. Finally, I present a biomechanical dataset combining mechanical properties of leaf lamina and mid-vein relative to the diverse leaf shape in genus Pelargonium. In chapter 4 and 5 on the one hand we investigate the volumetric leaf and midvein data and on the other hand generate in-field three-dimensional date set on diverse leaf shapes. In this further study we perform a series of mechanical measurements on leaves in a bid to dissect the possible structural correlation of lamina shape, mid-vein strength and leaf geometry. We use leaves of 17 morphologically diverse Pelargonium species and generate mechanical measurements that quantify the tissue strength relative to leaf morphology. Contrary to our expectation we find weak structural and morphological correlations, we also find weak correlation with material content of the leaf and it’s mechanical strength. Our results indicate that in genus Pelargonium a multivariate role of environmental, physiological and structural traits contributes to exhibited leaf morphology. As yet, studies integrating the three-dimensional leaf shape with relevant physiological and structural traits have received little attention. As a result there have been no overarching theory predicting patterns of variation in threedimensional leaf shapes and its relevance to leaf function. Fundamental questions around the underlying mechanisms of primary leaf functions such as leaf light interception and thermoregulation remain intrinquinly unresolved. Mostly limited due to flat (2D) surface representation of leaves, we suggest that further investigation will highlight the influence of three-dimensional leaf surface on intercepted light and thermal heterogeneity of leaves. This series of studies is a first framework for developing an interdisciplinary approach to address the multidisciplinary factors governing the immense diversity of angiosperm leaves. Further application of such techniques and investigations based on presented framework may allow us to map the laws of physical and biological crosstalk, which generates diverse leaf shapes.
dc.language.isoen_AU
dc.subjectLeaf form
dc.subjectMachine Vision
dc.subjectApplied optics
dc.subjectSfM
dc.titleOn growth and form of leaves in 3 dimensions: Applications of machine vision and advanced optics
dc.typeThesis (PhD)
local.contributor.supervisorNicotra, Adrienne
local.contributor.supervisorcontactadrienne.nicotra@anu.edu.au
dcterms.valid2019
local.description.notesthe author deposited 26/06/2019
local.type.degreeDoctor of Philosophy (PhD)
dc.date.issued2018
local.contributor.affiliationResearch School of Biology, The Australian National University
local.description.embargo2020-12-26
local.request.emailrepository.admin@anu.edu.au
local.request.nameDigital Theses
local.identifier.doi10.25911/5d134a1ae0def
local.mintdoimint
CollectionsRestricted Theses

Download

File Description SizeFormat  
Amit K. Singh Thesis 2019.pdf22.55 MBAdobe PDF    Request a copy


Items in Open Research are protected by copyright, with all rights reserved, unless otherwise indicated.

Updated:  22 January 2019/ Responsible Officer:  University Librarian/ Page Contact:  Library Systems & Web Coordinator