Lachhani, K; Duan, J; Baghsiahi, H; Willman, E; Selviah, DR; (2014) Error Metric for Indoor 3D Point Cloud Registration. In: Coleman, S and Gardiner, B and Kerr, D, (eds.) Proceedings: IMVIP 2014: 2014 Irish Machine Vision and Image Processing. (pp. pp. 34-42). IPRCS Irish Pattern Recognition and Classification Society: Derry-Londonderry, UK.

Text IMVIP 2014-2-2.pdf Download (528kB)

Abstract

An increase in commercial availability of 3D scanning technology has led to an increase of 3D perception for a variety of applications. High quality scanners require to be stationary and so multiple scans are required and subsequently need to be registered. A new error metric for registration based on the deviation of registered planar surfaces is introduced here and compared with a commonly used metric: mean square point-to-point distance. Four different sets of features are used to register six scans, the point-to-point errors are compared to the new error metric, planar surface deviation, and a disparity is observed for certain sets of features. The two metrics agree as to which sets of features gave the best registration but disagree as to which set produced the worst registration. It is concluded that further analysis and evaluation is required to determine which metric is more meaningful as a representative measure of registration accuracy and to also investigate other error metrics.

Download activity - last month

Export as JSONXMLCSV

Download activity - last 12 months

Export as JSONXMLCSV

Downloads by country - last 12 months

Export as CSVXMLJSON

Archive Staff Only

View Item