Geometric model for vision-based door detection

Abstract

Emerging assistive technologies, such as assistive domotics and socially assistive robots have considerable potential for enhancing the lives of many elderly and physically challenged people throughout the world. Blind and visually impaired people can use these technologies for many tasks recognizing objects, handling various household duties, navigation in indoor and outdoor environments. Door detection is one of the important issues in indoor navigation. This paper presents a novel vision-based door detection technique. It is based on the geometric properties of the 4-side polygon. The efficacy of the proposed method is tested using large database of images with different levels of complexity. The experimental results show the robustness of the proposed method against changes in colors, sizes, shapes, orientations, and textures of the door. Detection rate of 83% with relatively low false positive rate for simple images is achieved. This proposed algorithm is suitable for real-time, portable applications where it only requires one digital camera and low computational resources.