A prototype and algorithms for tree ring area measurement

Abstract

The analysis of tree cross-sections (discs) by measuring the rings (annuli) of a disc to determine a tree's growth history is conducted frequently by foresters, botanists and ecologists. The purpose of this paper is to discuss algorithms and approaches that were used to develop a prototype for a system to automate the measurement of tree rings through image analysis. A system was developed to define the area of annuli of a disc with clearly defined boundaries and no perturbations. The area of the entire region enclosed by a ring is determined by defining a ring's perimeter using a conventional radially-sweeping contour-tracing algorithm. The area of an annuli is calculated by subtraction of the area enclosed by the adjoining inner rings. This portion of the system is O((r+1)n) for the measurement of a disc, where n=(Ymax-Ymin+1) and r = the number of rings in the disc. A flood (polygon) fill algorithm must be used to verify the center of the disc was found. Thus, the efficiency of the entire system is O(n2). The area that must be visited by the flood fill algorithm is small thus the approach is efficient. Its utility for tree ring analysis is limited due to tree discs often having indistinct boundaries and flaws. Research is being conducted on image capture, edge detection and contour tracing methods to further develop the prototype. These capabilities must be incorporated to make this system a useful tool for analyzing a normal imperfect disc.