Automated Methods of Tree Boundary Extraction and Foliage Transparency Estimation from Digital Imagery

Foliage transparency in trees is an important indicator for forest health assessment. This paper helps advance transparency measurement research by presenting methods of automatic tree boundary extraction and foliage transparency estimation from digital images taken from the ground of open grown trees. Extraction of proper boundaries of tree crowns is the initial step in the determination of crown transparency. Subsequent processing methods are reliant on this boundary being correctly delineated. Here, image processing techniques are used to extract tree crown boundaries which are then approximated using spline curves. The resultant splines created by the automatic process can be modified by users in order to represent numerous boundary shapes caused by different crown conditions. In addition, spline coefficients that compactly represent the boundary can be stored for later retrieval, and therefore the system is capable of following changes in each tree over time. Once proper boundaries are extracted, the foliage transparency is estimated with three different methods. (i) a simple ratio method calculates transparency from a region ratio of foliage area to the inside boundary, (ii) a local method uses a fixed-size window to evaluate transparency for each small region, and (iii) a region tessellation method constructs a triangular mesh using a set of points inside and on the boundary and calculates transparency for each triangular region. While the first method gives us a classical transparency measure, the other methods provide a spatial distribution of foliage transparency. Experimental results show that all three methods provide reliable transparency estimates.