Mathematical Models for Chlorophyll Prediction in Maize Leaves

The paper presents mathematical models for chlorophyll prediction in maize leaves using digital images of the plants. Estimation of chlorophyll content is important to determine the phase of plant growth, which depends on what fertilization in cultivated areas will be needed, and to predict the period in which it will be required, which is a major part of precision agriculture. To evaluate the chlorophyll and carotenoids of the plants, digital images in RGB model and their transformation into HSV and Lab are used. A strong correlation between the color components and the three forms of chlorophyll - total, type a and type b - was evaluated and found. Three most informative features for modeling dependencies are selected - for total chlorophyll and chlorophyll type a – V(HSV), R(RGB) and S(HSV) and for chlorophyll type b – V(HSV), R(RGB) and H(HSV). The modeling of these dependencies with eight mathematical models in the Matlab environment is done. The fine tree type model is suitable for predicting total chlorophyll and chlorophyll type b, the coarse tree type model is suitable for predicting chlorophyll type a. To predict carotenoids, it is necessary to select other informative characters from color components or indices. The resulting models for predicting chlorophyll in maize leaves will be supplemented with two more factors influencing plant growth and how they are treated during the individual main growth phases - temperature and humidity - to predict how long the corresponding phase will occur and what it will be necessary to treat the plants in order to obtain a higher yield.