Tree ring area measurements using morphological image analysis

This paper describes a semiautomatic methodology for measuring the areas of annual tree rings. The cross section of a tree stem is first prepared by sanding and dipping into polyethylene glycol to prevent it from cracking. A digital grey-scale image of the resulting stem disk is then acquired by a scanner at the spatial resolution of 600 dots per inch. This image is processed by a series of morphological image processing transformations so as to automatically outline the tree rings. User interaction is restricted to the correction of extraneous or missing boundaries generated by disturbing features such as knots and low-contrast or very narrow rings. The methodology has been developed for Picea abies (L.) Karst., and its applicability to other species is discussed.