A Neural Network Based Technique for Staircase Detection using Smart Phone Images

The problem of stair case detection is important for various applications that include accident avoidance for elderly people and visually impaired persons. This also has significance in autonomous robotic systems that requires a stairs detection module. The proposed system addresses the problem of stair case detection using handheld mobile devices. The algorithm works using images captured in a mobile phone. It uses both image and geo-spatial features. The performance of individual features in the context of stair detection is studied using a neural network based system. The obtained results with the proposed algorithm is explained in the paper.