Research Reports

• Research Reports Home
• UC Research Repository
• Honours Reports
• Reports Manager (CSSE Staff Only)

Abstract for HONS 05/20

Registration of 3D scans taken from opposite sides of a vineyard row

Campbell Mercer-Butcher
Department of Computer Science and Software Engineering
University of Canterbury

Abstract

The automation of data collection on the crops in vineyard environments can lead to increased efficiency and accuracy, and decreased costs. One step in the pipeline of automation is the registration of scans taken from opposing viewpoints of the vineyard rows. Whilst the single side reconstructions can be obtained with traditional mapping techniques, merging two opposing sides can not be achieved with standard registration algorithms. We propose a methodology of using global static feature segmentation as the basis of registration. Our quantitative evaluation across multiple virtual datasets, demonstrated the accuracy of our method.