Accurate area determination of complex leaves using digital image analysis

Abstract

A recently described computer image analysis method was modified for quantification of leaf area. These areas were compared to those estimated by a lower resolution method based on planimetry. Leaves of Eucalyptus nitens (Deane & Maiden) Maiden, pinnae of Dicksonia antarctica Labill. and leaves of Acacia dealbata Link were used for analysis, offering surfaces of low, intermediate and high complexity, respectively. Low-resolution planimetry was found to be a suitable method for the calculation of leaf area of simple broadleaves. However, for surfaces of greater complexity, the higher resolution of image analysis gave more accurate estimates of area. Overlapping of primary pinnae in the complex A. dealbata leaf proved to be a larger source of error than inadequacy of resolution.