In this article, we have considered the young apple trees leaf area measurement. As you will find, we conducted the analysis of the three measurement methods of leaf blade area of young bi-axis apple trees on rootstock 54-118. As a result, we established high accuracy of the measurement results. Particularly, by a method of area calculation within a leaf contour on millimetre graph paper the area was 31,21 sq. cm. And with help of a method of Computer Vision using application “Petiole” on smartphone with operational system Android the area, which we obtained, was 31,28 sq. cm. The gravimetric method and the use of mobile application “Petiole” appeared to be the most productive as to time consumption to carry out the operation.
Key words: leaf blade area, definition methods, Computer Vision, millimetre graph paper method, gravimetric method, “Petiole” application, bi-axis apple tree.
Authors: O. V. Polunina, Post-graduate Student, Uman National University of Horticulture (Uman), Ukraine V. P. Maiboroda, PhD in Agricultural Sciences, Associate Professor, Uman National University of Horticulture (Uman), Ukraine A. Y. Seleznov, Chief Technology Officer at Petiole LTD, Ukraine
Photosynthesis products amount to 90-95% of dry substances used by a plant to build all its organs (including yield) [Fridrih et al. 1983, Kondratenko 1997]. A leaf is one of the main assimilating organs of a fruit plant, the formation of which depends on a complex of influence factors – nutritive and water regime, planting density, lighting, crown shape, application of protection means and plant growth regulators and many others [Fridrih et al. 1983, Mohsenin 1986]. Consequently, leaf blade area serves as a certain indicator of a plant condition as well as its potential productivity. Up-to-date means are to guarantee the accuracy of the measurement results, which is why new methods of the leaf blade area estimation appear, and this encourages a researcher, first of all, to analyze and estimate them.
The methods of leaf blade area estimation, which are widely used in agricultural research of fruit crops, require considerable amount of time to do the measurement or they are allowable to a high error value [Klochkova et al. 2010].
In 1971 the following methods of leaf area estimation were described and analyzed by a group of Check scientists: leaf area calculation within their contours on millimeter graph paper, planimetric, photometric, use of linear leaf parameters, gravimetric, etc. [Sestak et al. 1971]. In particular, the difference of the methods as to their complexity, accuracy and productivity was stated [Moiseichenko & Yeshchenko 1994].
Drawing a leaf contour on millimetre graph paper is one of the most accurate methods of leaf blade area estimation [Pandey & Sing 2011], but its productivity is low [Moiseichenko & Yeshchenko 1994]. Gravimetric “carvings” method, which is widely used for the majority of agricultural crops, is characterized with higher productivity [Moiseichenko & Yeshchenko 1994]. The accuracy of this method depends on many factors: a chosen place on a leaf influences carving thickness; veins in the chosen carving place increases its mass; fast leaf shrivelling results in mass loss and others [Klochkova et al. 2010].
The methods using the programs which are based on the analysis of scanned leaf pictures have become more common with the development of computer technologies [Klochkova et al. 2010, Nurminskaia et al. 2017]. These measurement methods are efficient for the crops with a leaf blade area which does not exceed a scanner size.
Ukrainian designer Andrii Seleznov suggested a modern method of leaf blade area estimation with help of “Petiole” application using a library of Computer Vision Open CV for Smartphone of Android system, which allows receiving data from leaf blade area both in the laboratory and in the orchard or the nursery in the regime of real time. In fact, the development of new methods and the improvement of existing ones in the research in fruit production and in nursery studies. In particular, the aim of one of them is to enhance the productivity of a scientific process at an experimental level. First of all, due to time-saving, spent on getting the results, and the accuracy increase of the data, which is received. Besides, an important requirement for a present-day researcher is the mobility. This factor is very much favoured by the development of modern technologies.
We conducted our research at the Department of Fruit Growing and Viticulture. Our location was an experimental fruit nursery plot of Uman National University of Horticulture (Ukraine).
We had a task to analyze its accuracy and efficiency of leaf area measurement, in comparison with conventional methods. For example, we focused on area calculation within leaf contours on graph paper and gravimetric method. At the same time, we use one more new method for this comparison. It is a digital instrument “Petiole” mobile application on a Smartphone – for leaf blade area estimation,
The analysis of the received data was made in the context of the experiment with growing annual young bi-axis apple trees, cv. Florina, on root stock 54-188, depending on grafting height and the way two axis were formed.
Overall a two-factor trial has eight variants in a four-time replication. Basically, we took a sampling of ten apple trees leaves from each experimental plot. Totally, 32 samplings all together. We measure an area of each leaf with these three methods in succession.
First of all, we took the method of area calculation within leaf contours on millimetre graph paper as the control. As the method allows saving leaf wholeness, that is why we use this method firstly.
Then, we used the method of Computer Vision measurement with mobile applcation “Petiole” to estimate leaf blade area. The use of this method envisaged the following actions.
Finally, we used a gravimetric method, as it implied leaf blade damage. We got the data by this method following the next steps of the processes. Firstly, we chose leaves in samplings, at least 10 leaves each. Then we cut off petioles on each leaf. As a final step – we checked the weigh of each leaf blade.
Together with our measurements, we took 20 carvings. We made holes with help of a drill on both sides of the central vein. We got them from a sampling of leaves, and later they were also weighed [Karpenchuk & Melnik 1987].
Leaf blade area calculation is based on an equation [Moiseichenko & Yeshchenko 1994]: Petiole leaf area measurement
S – leaf blade area, sq. cm; Sj – area of one carving (S1=0,785 x D2, where D – carving diameter, cm), cm2; n – number of carvings, pieces; M – mass of leaves in a sampling, g; m – carving mass, g; and N – number of leaves in a sampling, pieces.
As a result, the use of mobile application “Petiole” and “carvings” method helped to reduce the time for leaf blade area estimation. In comparison with millimetre graph paper method it is more time consuming (Table 1).
[table id=1 /]
In comparison with other methods, the use of “Petiole” application allowed minimising an accuracy error of the received data due to the reduction of the effect of human factor on the measurement process. Thus, according to the results of leaf measurement of Florina apple trees on rootstock 54-118, we received the value, the closest to the control, with the use of “Petiole” application (Table 2). The deviation from the values of the control method was 0.08 sq. cm or 0.22 %. Also, gravimetric “carvings” method showed a serious deviation from the control, namely, 0.80 cm2 or 2.56%.
[table id=2 /]
Significantly, we prove relationship. In the light of this finding, relationships between leaf area estimated by gravimetric “carvings” method (dependent variable) and that estimated by millimetre graph paper method (independent variable) for thirty-two samplings as given in Figure 1 (B) show that the two sets of estimates are strongly related with each (R2 = 0.829).
Particularly, the closest relationships between leaf area estimated by “Petiole” application and leaf area estimated by millimetre graph paper method indicates the high accuracy of measurement using “Petiole” application (Figure 1 (A)). In the light of this, the coefficient of determination between the two sets of estimates prove that relationships are linear and significant (R2 = 0.994). Figure 1: Relationship between leaf area estimated by millimetre graph paper method for apple treesFigure 1: Relationship between leaf area of young apple trees estimated by millimeter graph paper method and: A – leaf area estimated by “Petiole” application; B – leaf area estimated by gravimetric “carvings” method.
Evidently, the method of use of “Petiole” application is more accurate, faster and nondestructive (saves leaf wholeness) and can be applied in nursery plot or orchard. An important factor which affect the accuracy of use of “Petiole” application is uniform lighting of the leaf blade under the camera so that there are no shadows that affect the data error.
Accordingly, we found proof of the effectiveness and accuracy of the new method of leaf blade area estimation with the results of this research. With help of Computer Vision a smartphone camera as well as “Petiole” mobile application helps to get indicator values more efficiently and accurately. By comparison with the other methods, it decreases the number of manual manipulations and also reduces measurement time. The application of this method makes it possible for a researcher to improve labor productivity during measurement processes.
The authors thank Professor O.V. Melnyk for research assistance.
Original Text of the article “Evaluation Methods Of Estimation of Young Apple Trees Leaf Area”: here