Computer Modeling of Fiducial Markers Images to Estimate the Accuracy of Object Positioning in Space

Abstract

Fiducial markers are widely used in machine vision systems. One of the main ways of their application is to estimate the relative position of the camera and marker in space. The latter are visual artificial objects, which are designed for their robust detection among other objects in the environment of the images. Examples of fiducial marker accuracy estimation are found in many studies. Usually such studies are conducted for one or two or three code marker systems. And accuracy estimates are given only for specific parameters of the machine vision system, which is designed by the authors of the work. This approach is not universal, and for other parameters it is necessary to carry out the error estimation again. This paper presents an approach to determining the error for comparing more fiducial marker systems with each other using more universal metrics. For this purpose, computer simulations of marker images were performed for five different systems with different spatial orientations. The images were processed using detection algorithms proposed by the authors of the marker systems. The processing results were compared with the specified parameters in the simulations. Relative error estimates were obtained for the five different code marker systems.