Extraktion physiologischer Koordinatensysteme von Pflanzenwurzeln und -blättern aus Bildsequenzen

The present work develops a method for analyzing the growth of plant leaves and roots in natural object coordinates using image processing techniques. The method is based on the structure tensor method that has already been used for the acquisition of growth charts in image coordinates. The main advantages of growth measurements in physiological coordinates is the possibility to compare the results of different measurements and the direct interpretation of the results. A method for extracting the coordinate axes is developed both for leaves and for roots. The center line of a root is identified as the physiological coordinate axis by means of a method based on active contours. In the case of leaves, the physiological coordinate axes are represented by their veins. The latter are searched for by use of a tracking algorithm which relies on matching methods. All physiological coordinate axes are represented in the form of B-splines. The growth charts are sampled from the B-spline positions, thus providing a coordinate transformation into the corresponding physiological coordinate system. The transformation has subpixel accuracy. Thus, it significantly stays behind the spatial resolution of the growth charts. The accomplishment of the present method provides a powerful tool for growth analysis of plant leaves and roots.