Leaf Area Determination through Mathematical Models Based On Image Analysis.

Total leaf area (TLA) is one of the most relevant features to estimate yield and productivity in horticultural plants. Direct methods used to estimate TLA cause damages to the plants and are time consuming. Indirect methods, such as image analysis, demand accuracy in the set up of the measurement procedure, sometimes very difficult to achieve. Computer tools might be the solution to search for new indirect and reliable approaches for producers and researchers in this task. The aim of this work was to build mathematical models based on indirect measurements to estimate TLA in coffee plants. These indirect measurements were image analysis, light intensity and leaf area index (LAI) obtained via LAI-2000 system. Total leaf area data obtained by using destructive method (Li-3100c) were used as a reference. Coffee plants, cultivar Topazio, with a canopy ranging from 0.43 to 1.14 m height were used in the experiments performed at Federal University of Lavras, Brazil. Two models were developed based on image analysis and on measurements of canopy height, width and their projected area. Also, a fisheye lens adapted to a commercial digital camera was used to produce one image taken above the plant and another four taken from below, when other two models were built. In addition, three other models were also developed via luximeter data measurements and via LAI-2000 system. Comparing the models, the results showed an R² of 0.94 for the projected area model, an R²  of 0.82 for plant height and width model and an R² of 0.83 for the fisheye lens model (above image). The results suggest that these models can be used to estimate TLA in coffee plants. The other models, based on the luximeter and on the LAI 2000, presented low R² values, and further studies are necessary until they can be properly used.