| Hi-index | 0.00 |
Improving plant productivity is key to address major challenges such as global food security. Plant phenotyping is an emerging science for these challenges and is the study of the dynamic interaction between genotype and environment. It is well known that plant temperature is an indicator of plant stomatal conductance, which is linked to the process of nutrient uptake and consequently plant growth rates and yields. Thermal images can be used to measure temperature of individual plants and their differences under stresses. Therefore, thermal image analysis can provide important clues for genetic screening. In order to accurately measure plant temperatures and their differences, we have to remove the impacts from background thermal inhomogeneities. To do so, we propose a novel algorithm to estimate the background using surface fitting with RANSAC algorithm. We also designed the experiment to combat the air temperature changes during imaging periods. Experimental results have shown that the proposed algorithm is able to significantly reduce background inhomogeneities in infra-red images to accurately estimate temperature differences between background and plants.