In: Konstantin G. Korotkov (Ed.). Measuring Energy Fields: Current Research. – Backbone Publishing Co. Fair Lawn, USA, 2004. pp. 143-156.
Analyzing Coronas of Fruits and Leaves
Aleksander Sadikov, Igor Kononenko, Franco Weibel*
University of Ljubljana,
Faculty of Computer and Information Science, Trzaska 25, 1000 Ljubljana, Slovenia e-mail: {aleksander.sadikov; igor.kononenko}@fri.uni-lj.si
* Research Institute of Organic Agriculture (FiBL); 5070 Frick, Switzerland e-mail: franco.weibel@fibl.ch
Abstract
We implemented a system GDV Assistant for parameterization and visualization of coronas of humans and plants. Besides standard parameters, developed by the team of prof. Korotkov, our program includes some additional numerical parameters. In last few years in several studies we recorded coronas of apple tree leaves and fruits in order to verify and compare their vitality under different conditions. We used GDV Assistant for preprocessing and for numerical parameterization of coronas and we used various machine learning algorithms for analyzing the databases of parameterized corona pictures. The results of our studies show that coronas of leaves and fruits give useful information about the stress status of plants and about the variety. However, we were not able to differentiate between organically and conventionally grown fruit which were similar in their standard quality parameters such as fruit flesh firmness and sugar content.
1 Introduction
Recently developed technology, based on Kirlian effect, for recording the human/plant bioelectromagnetic field using the Gas Discharge Visualization (GDV) technique provides potentially useful information about the biophysical and/or psychical state of the object/person (Korotkov, 1998). The recorded coronas are then processed with GDV Analysis software and described by the set of numerical parameters. In the previous study (Skocaj et al., 2000) we recorded coronas of grape berries and have shown that standard numerical parameters of coronas can be used to successfully classify berries according to infection and sort.
In studies, described in this paper, we were interested in vitality of plants in various stress status (healthy versus infected plants), different varieties, different rootstocks and grown under different systems (organic versus conventional; various fertilization methods). The recordings were done at the Institute for Organic Agriculture FiBL at Frick, Switzerland. In order to improve the parameterization of pictures of coronas we developed a system GDV Assistant (Sadikov, 2002) which implements several additional numerical parameters for describing coronas of human fingers and plants. We used several different machine learning algorithms for analyzing the parameterized coronas. For an introduction to machine learning paradigm see for example (Mitchell, 1997).
The paper is organized as follows. We start with description of new parameters, introduced by the GDV Assistant system. We follow by describing our recording methodology for obtaining coronas of plants. Section 4 describes various classification problems obtained by recording coronas of plants under various scenarios. Section 5 briefly describes machine learning algorithms used in our study and Section 6 provides results of the analysis. Finally, in Section 7 we conclude and give some ideas for further work.
2 GDV Assistant
GDV Assistant (Sadikov, 2002) was implemented in order to allow more flexible analysis of coronas than provided by standard GDV software suite (Korotkov, 1998). We used the first nine numerical parameters, as returned by GDV Analysis: A1. Area of GDV-gram, A2. Noise, deleted from the picture, A3. Form coefficient I, A4. Fractal dimension, A5. Brightness coefficient, A6. Brightness deviation, A7. Number of separated fragments in the image, A8. Average area per fragment, A9. Deviation of fragments’ areas. We used also two parameters, defined by Korotkov and Korotkin (2001): average streamer width and entropy of corona. These parameters are a reimplementation of those in the original GDV software suite, therefore their exact values are usually somewhat different. Besides we defined four additional parameters:
1. form deviation;
2. normalized skewness of brightness;
3. normalized stability of brightness;
4. entropy of brightness.
We also implemented and used seven parameters developed by Hu (1962).
Form deviation (FDev) is similar to Form coefficient I. It is also defined on the basis of curves of constant luminosity (isolines) that are defined in (Korotkov and Korotkin, 2001). We created this parameter to emphasize the important changes in corona’s form even more than Form coefficient already does.
The formula is:
where F[n] is distance between the center of corona and the n-th point on the isoline and avgF is the average distance between the center of the corona and the isoline.
To further extract the information contained in the corona’s histogram we defined three additional parameters based on it. These are Brightness skewness (ν3), Brightness stability (ν4) and Brightness entropy (H) and they respectively give us information on the slope, stability and uniformity of frequency distribution of corona’s brightness. Respectively, the formulas are: