Colloquium "An Improvement to Least Squares Line Fitting Based on Gradient Orientation"

Abstract:
Least squares line fitting is one of commonly used methods in the area of computer and robot vision. We propose a novel method for least squares line fitting that uses both the point coordinates and the local gradient orientation to fit an optimal line by minimizing the proposed algebraic distance. The proposed inclusion of gradient orientation offers several advantages: (a) one data point is sufficient for the line fit, (b) for the same number of points the fit is more precise due to inclusion of gradient orientation, and (c) outliers can be rejected based on the gradient orientation or the distance to line. In this colloquium we will give an overview of standard methods for least squares fitting of lines, circles and ellipses that use only point coordinates. We will show how to extend the least squares line fitting to include both point coordinates and gradient orientation. We will discuss how it may be done for circles and ellipses, which is our current research topic.

CV:
Tomislav Petković received his Dipl. Eng. in Electrical Engineering in 2002, his MSc in Electronics in 2006, and his PhD in Electronics in 2010, all from the University of Zagreb, Croatia. From 2002 till 2013 he was a research and teaching assistant at the Department of Electronic Systems and Information Processing at the Faculty of Electrical Engineering and Computing, University of Zagreb. From 2013 he is a researcher on ACROSS FP7 project. His main field of interest is digital image processing and analysis, in particular biomedical and industrial applications.