Accurately Combining Skeletal Models
Abstract
Skeletal tracings are a compact and efficient way of storing large amounts of data and are particularly useful for analysis in the world of biological data. One of the largest issues with skeletal tracings is that they are subject to so many dependent variables that, in most cases, it is impossible to receive a perfect representation from the initial data set. A possible solution for more accurately representing a given data set would be to take multiple tracings of the same data, produced by different tracing algorithms or with altering parameters, and perform an analysis on these sets to create an average model. In the past few years, there have been a few advancements concerning combining skeletal tracings, but none of them have concluded with an order independent solution. This is the key feature that has been the driving factor in this research, because without the resultant model being order independent, the method could return different resulting models for the same set of data. In this document, I propose a method for analyzing and averaging a given set of skeletal tracings that will result in a merged model that will be a more accurate representation of the original data set. This method uses a new technique for K-Means clustering of polylines to analyze the geometric similarity and explores a clustering and merging technique for specified points to merge the models topologically. After combining the models, I then propose a way of visually analyzing the accuracies associated with the different techniques.
Citation
Witt, Luke Dakota (2020). Accurately Combining Skeletal Models. Undergraduate Research Scholars Program. Available electronically from https : / /hdl .handle .net /1969 .1 /175402.