This paper proposes objects tracking method by association with RFID and visual sensors. The detection from RFID readers with proximity to each object results in coarse localization. The estimation is determined by a virtual sensor which is comprised of RFID readers detection combination. Each virtual sensor has a reference point which represents the center of possible positions. After the coarse estimation, image frames from two cameras improve the estimation close to a real position based on parallel projection model. The visual compensation sufficiently accomplishes the refined tracking. The experimental results show our proposed algorithm remarkably accomplishes the localization performance.