Automated Quantitative Analysis of Terminal Tree Branch Similarity by 3D Registration

A popular, but unsubstantiated view is that tree branch morphologies are similar (self-similarity) and are of an iterative nature. To date there are no studies that document plant branch self-similarities. The purpose of this research is to develop a program (3D Simquant) that estimated self-similarities among paired branch terminals quantitatively. After 3D Simquant was written, the program was verified and sensitivity analysis performed, eighty-five terminal branch pair-wise comparisons from five different tree species were analyzed. Only two branch geometries (Y and Y+1 terminals) were compared. Simple Y terminals are terminal main stems with one side branch while Y+1 terminals are main steams with two side branches. Similarities among paired branch terminals were quantified with Root-Mean-Square-Error (RMSE) after registration of images. For the five species tested, all Y terminals had RMSE values less than 1.5 which indicates they were similar. For most Y+1 terminals, RMSE values were twice that of Y terminals indicating the Y+1 samples were more dissimilar than Y terminals. Overall, the programs were accurate and rapid for an analysis of branch similarities.