Outdoor computer vision and weed control / Udendørs Computer Vision og ukrudtskontrol

The PhD project investigates the potential outdoor support of computer vision within weed control for reduced use of herbicides. The use of computer vision outdoors encounters the problem that the illumination conditions may change from direct sunshine to a clear blue sky as the illumination source, a change that radically alters the colours and conditions for the image formation. The results of the thesis demonstrates that a priori knowledge and modelling of the image formation process may be of significant importance for development of image analysis methods and sensors for robust analysis of colour information captured under varying illumination conditions. This opens for general perspectives for development of computer vision methods and sensors for outdoor use. The results of the project are now turned into the first commercial automatic row guidance system manufactured by the company Eco-Dan A/S and sold in Europe and North America. Among many prizes the product received the European IST grand prize 2002. The activity with outdoor computer vision and analysis of plant growth is continued through a post-doc project investigating the potential use of computer vision for detection of nutrient deficiency and potential uptake. Funded by Danish Research Councils. (Hans Jørgen Andersen, Erik Granum)