Dynamic background modeling for a safe road design

摘要: In this paper we propose a new dynamic algorithm able to automatically identify the background from images under severe changes of environmental conditions. The exploits motion estimation techniques and subtraction methods. proposed scheme is implemented in road construction application scenarios outdoor environments which camera vision dramatically varied with respect weather Despite such variations, can be detected improving tracking performance for vehicles just assisting survey engineers defining possible errors design yielding as deviations between actual ideal trajectory vehicle over turn. Experimental results real-life surveillance systems reveal efficiency regardless significant changes. addition, requires low computational complexity.