A diagnostic tool for magnesium nutrition in maize based on image analysis of different leaf sections

Abstract

The nutritional status of maize (Zea mays L.) can be diagnosed by chemical analysis of leaves, which is very slow, or by visual diagnosis of deficiency symptoms, which is dependent on observer experience. The artificial visual system (AVS) is a technology to identify nutritional deficiencies in maize, allowing correction for nutrient supply at earlier development stages in maize. Our objective was to propose methods of artificial vision and pattern recognition to identify the concentration of magnesium (Mg) in maize plants grown in the greenhouse. Magnesium concentrations were 0.0, 0.65, 1.3, and 2.0 mM Mg, with four replications. Leaf scans were collected at V4 (four leaves fully developed), V6 (six leaves fully developed), and V8 (eight leaves fully developed) stages, and these leaves were samples for chemical assays. Such images were processed using AVS methods. Volumetric fractal dimension (VFD), Gabor wavelet (GW), and VFD with canonical analysis (VFDCA) were techniques used by the AVS to extract deficiency characteristics in the leaf images. The increase of Mg in the nutrient solution caused an increase in the Mg concentration in leaves, resulting in typical visual symptoms. The AVS method was able to identify all levels of deficiency, scoring 75.5% of rights in images of the middle section of leaves in the VFDCA method, in color scans of V4 leaves. The AVS was efficient at diagnosing Mg concentrations in leaves of maize during the V4 stage.