Comparision of focus measures for collision avoidance application of mobile robots

Abstract

Low cost fixed focus cameras are focused to infinity. This will introduce an amount of blur in the image of objects in the close range. This blur effect is called defocusing effect. Researchers in the past introduced focus measures to quantify the defocusing effect in an image. Our purpose is to find the suitability of Focus Measures for collision Avoidance Application of mobile robots. In order to do that, an experimental setup was developed with a low cost fixed focused camera, single board computer and an ultrasound module. Controlled experimental data was collected using the developed setup. Using the collected data, four different focus measures from the literature were analyzed for three different field of views. Results show that GLVN focus measure in the full field of view is more suitable for the collision avoidance application.