Extraction and Distortion Rectification of Signboards in a Scene Image for Robot Navigation

Abstract

In this paper, we describe an approach for finding signboard information in a grey scene image and rectifying a distorted signboard image taken from a nonorthogonal viewing direction. First, signboards in scene images are detected using several heuristics of characters and character lines. Then, we attempt to compute the orientation of the camera relative to a detected signboard from a single view. Finally, we rectify the distortion of signboard images, making the recognition of characters under a wider range of viewing conditions much easier. We show experimental results that demonstrate the effectiveness of the proposed approach. Lastly, we present an application of the proposed method to robot navigation using signboards on walls and doors as landmarks, where we could navigate a mobile robot in indoor environments successfully. Using signboards in scenes as landmarks enables us to realize humanlike navigation and map generation, and also benefits human-robot communication.