Scanner image methodology (SIM) to measure dimensions of leaves for agronomical applications

Abstract

A scanner image methodology was used to determine plant dimensions, such as leaf area, length and width. The values obtained using SIM were compared with those recorded by the LI-COR leaf area meter. Bias, linearity, reproducibility and repeatability (R&R) were evaluated for SIM. Different groups of leaves were scanned and measured. R&R studies showed that, the lowest SIM’s resolution was nineteen categories. SIM’s R&R ANOVA showed the method’s measurement error was not significant. In the image processing method, the color image was converted to gray scale over the green band and it was segmented using Otsu methodology. The noise produced was cleaned with a median filter. The leaf image was rotated to align the longest parallel line to y-axis or x-axis using central moments. From the centroid using directional erosions the leaf width and length were obtained and recorded.

Key words: Leaf area, width, length, digital image analysis, segmentation.