Bisector-Based Tracking of In Plane Subpixel Translations and Rotations

Abstract

We present a method for distance measuring planar displacements and rotations with image processing methods. The method is based on tracking the intersection of two non-parallel straight segments extracted from a scene. This kind of target can be easily identified in civil structures or in industrial elements or machines. Therefore, our method is suitable for measuring the displacement in some parts of structures and therefore for determining their stress state. We have evaluated the accuracy of our proposal through a computational simulation and validated the method through two lab experiments. We obtained a theoretical mean subpixel accuracy of 0.03 px for the position and 0.02 degrees for the orientation, whereas the practical accuracies were 0.1 px and 0.04 degrees, respectively. One presented lab application deals with the tracking of an object attached to a rotation stage motor in order to characterize the dynamic of the stage, and another application is addressed to the noncontact assessment of the bending and torsional process of a steel beam subjected to load. The method is simple, easy to implement, and widely applicable.

Keywords

• object tracking
• image processing
• subpixel accuracy
• deflection dynamics