Support vector machine for overcoming the problem of vanishing point during stairways detection

Stairways region detection from a stairways image is a challenging activity to support autonomous system and visually impaired people for navigating safely. This paper proposes a framework for investigating the problem of vanishing point during stairways candidate region verification. For that, initially stairways candidate region is detected utilizing the unique natural and geometrical features of stairways. One unique natural feature is three connected point (3CP). This 3CP is formed at every stairways step's horizontal edges end points with stairways step's width and height edge's intersection point. Another geometrical feature is stairways step's edges are exhibited in sorted order. These geometrical features are used to detect stairways candidate region from stairways images. Where, the 3CP geometrical feature of stairways is used to validate the stairways step's horizontal edge segments. After that, the validated stairways edges are justified by computing vanishing point (VP). This justification ensures that the edge segments are arrived in increasing parallel order. Finally, the y coordinate value of VP is utilized to verify the edge segments of stairways from other analogous looking objects and ensure the detection of stairways region. However, in some cases the vanishing point does not distinguish the stairways region from other stairways like objects such as zebra-crossing or rail-line. This paper investigates this problem and verified the stairways candidate region using the SVM classifier where Gabor filters are used to extract the features. Various stair images are utilized to evaluate the proposed framework and presented outcomes demonstrate the adequacy.