Automatic classification of leaf diseases using multispectral and stereo images
pattern recognition, classification of leaf diseases, feature extraction, sugar beet
The goal of this project is the development of methods for the automatic classification of leaf diseases based on high resolution multispectral and stereo images. As our exemplary domain we use leaves of sugar beet which may be infected by several diseases, such as rusts (Uromyces betae), powdery mildew (Erysiphe betae) and other leaf spot diseases (Cercospora and Ramularia beticola). Leaf diseases are economically important as they could cause a yield loss. Early and reliable detection and classification of leaf diseases therefore is of utmost practical relevance. High resolution images allow to identify the pathogen based on visible symptoms, such as change of colour and texture. Especially the multispectral appearance, the form and distribution pattern of spots and their evolution over time will be decisive. In order to be allow in vito analysis we need to take the 3D-pose and form of the leaves into account. In a first step we analyse single leaves and train the classifiers using healthy and intentionally infected leaves. Later we will analyse complete plants.
ACTIVITIES AND INFORMATIONS