Figure 2

Example measurements of lesion characteristics in Zea mays using imagery and computer algorithms. Many useful phenotypic attributes can be quantified via calculations from phenotype imagery. Here, we show a variety of phenotype appearances in maize as well as corresponding measurements to demonstrate the capability of algorithms in differentiating these appearances. Each section of the figure contains a series of images that vary by some characteristic as well as a chart showing the differences in the measurements (features) of this characteristic from the images. The top left section contains images of leaves that are (a) green, (b) yellow, and (c) brown. The color differences are clearly visible in the (d) chart of corresponding histograms based on the hue of the leaves. Similarly, the top right section contains leaves with lesions of varying shape: (e) tiny round lesions, (f) elliptical lesions, and (g) irregularly shaped lesions. A (h) histogram of lesion roundness shows how shape can be measured in a way to differentiate these lesion shapes. In the middle left section, leaves with (i) small, (j) medium, and (k) large lesions are shown, along with (l) a histogram of lesion size. The right middle section attempts to capture lesion distribution in leaves with lesions ranging from sparse to dense (m)-(o). An algorithm that plots a (p) histogram of the distance to the nearest neighboring lesion is used to provide a measure of lesion distribution. Finally, the bottom section contains (q)-(s) leaves with lesions in various spatial arrangements. By vertically partitioning a leaf into sections of equal width, the spatial configuration of lesions can be measured by counting the number of lesions in each partition, and this histogram is shown in (t).