3D Information retrieval in Mobile Robot Vision based on spherical compound eye

Three-dimensional scene information retrieval has became an important research topic in the subject of Mobile Robot Vision and is useful for many applications, such as Mobile Robot Navigation and Mobile Robot target tracking. This paper presents a 3D information retrieval system for Mobile Robot Vision, which relies on the spherical compound eye structure to get a spherical camera image acquisition platform. Novelties of our work are the spherical camera platform for image acquisition and the 3D information retrieval methods. The spherical camera image acquisition platform, which is based on the spherical compound eye structure, has advantages of a large field of view, multi-channel and high sensitivity. The 3D information is recovered by image feature extraction and tracking, information restored and surface optimization. The paper uses the SIFT algorithm, the ANN algorithm and the Ransac algorithm to extract and track features, adopts the SfM method and the PMVS method to get 3D scene information. In additional, the Poisson surface information retrieval method is adopted to optimize the 3D information retrieval result. Experiments results prove a great advantage of real-time and accuracy compared with other systems.