A vision-based approach for nondestructive leaf area estimation

Leaf area of a plant is an important indicator to analyze and assess the growth and estimate the production potential for the plant species under study. Leaf area measurements are important in estimating biomass productivity, adaptation to the environment, nutrition and soil-water relations. In this paper we propose a rapid, accurate and yet simple approach to estimate the leaf area using computer vision approach. The approach is based on analyzing images of plant leaves collected using an off-the-shelf digital camera. Images of a leaf is processed to extract the contour, then a transformation is applied to compensate for the projective distortion and finally the leaf measurements are calculated. The results are compared with measurements using grid-counting method.