Surface reconstruction of wheat leaf morphology from three-dimensional scanned data

Abstract

Realistic virtual models of leaf surfaces are important for several applications in the plant sciences, such as modelling agrichemical spray droplet movement and spreading on the surface. In this context, the virtual surfaces are required to be smooth enough to facilitate the use of the mathematical equations that govern the motion of the droplet. Although an effective approach is to apply discrete smoothing D2-spline algorithms to reconstruct the leaf surfaces from three-dimensional scanned data, difficulties arise when dealing with wheat (Triticum aestivum L.) leaves, which tend to twist and bend. To overcome this topological difficulty, we develop a parameterisation technique that rotates and translates the original data, allowing the surface to be fitted using the discrete smoothing D2-spline methods in the new parameter space. Our algorithm uses finite element methods to represent the surface as a linear combination of compactly supported shape functions. Numerical results confirm that the parameterisation, along with the use of discrete smoothing D2-spline techniques, produces realistic virtual representations of wheat leaves.

Study of retention variability on an early growth stage herbaceous plant using a 3D virtual spraying model

“…The reconstructed virtual plant was not watertight, which means that holes were not closed during the fusion of the different views. The scanning procedure provided high levels of microsurface details that were not essential in this study since they resulted in undesired surface orientation gradients (Kempthorne et al, 2015) that may skew the droplet incidence angle computation. In addition of this, a high density mesh required a higher computational time for the spray droplet interception algorithm (described below).…”

“…In consequence, the computed retention rates will also be smaller, which was not problematic from a comparative point of view of the simulations. Nevertheless, improvements of the surface reconstruction could be reached using the recent approach proposed by Kempthorne et al (2015), which guarantees surface reconstruction with continuous gradient. Another interesting approach could be based on the Lindenmayer system (Prusinkiewicz and Lindenmayer, 1990) to provide a well characterized plant model that could be used as a standard for comparing existing spray retention models.…”

Sensors for measuring plant phenotyping: A review

“…This system integrated lidar and other optical sensors to measure leaf in indoor conditions. In terms of field measurement, Gebbers et al [70] Figure 5 Leaf parameters measurement by laser sensors [65] 3.2.5 Comparison and analysis of leaf parameters measurements The above mentioned measurement methods can be divided into two types according to the measurement principle of leaf. Color digital camera, stereo vision system, range camera, and lidar can measure or estimate leaf features by segmenting and reconstructing the leaf and calculating its actual values.…”

From image processing to computer vision: plant imaging grows up

“…The combination of existing image processing and data clustering approaches, adapted to work with plant image data, provides an encouraging correlation with manually observed data, which can be obtained automatically on a high throughput phenotyping installation. Kempthorne et al (2015) address the problem of accurately reconstructing the shapes of leaf surfaces for use in modelling agrichemical spraying of whole plants, addressing the requirement of high-accuracy rather than high-throughput for phenotyping. Challenges produced by leaves bending and twisting, which complicates the application of traditional algorithms, were also investigated.…”

“…The approaches themselves can be mathematically founded, high-level approaches, which can be applied to a dataset from the top down; that is, how the data was captured is less important than the type of data it results in (for example, the 3D leaf modelling work of Kempthorne et al (2015)). Or they could be a combination of modified existing, bottom-up processing components which produce a robust and reliable measure, and can be implemented in a high-throughput manner in a phenotyping centre.…”