Research On Indoor Monocular Vision Localization Method Based On Artificial Landmark

With the development of the robot industry,indoor service robots have become the focus of research at home and abroad in recent years.Self-positioning technology is the key technology that indoor mobile robots need to solve firstly,because GPS and other global positioning systems cannot achieve accurate indoor positioning.Mobile robots need to use other sensors to get information about their surroundings,thereby achieving accurate indoor positioning of the mobile robots.This paper designs a monocular vision positioning system based on visual road signs,the problem of missing distance information in monocular visual is solved.In a complex indoor environment,the mobile robot self-positioning under the motion field of view is realized.It not only ensures high positioning accuracy,but also has strong operability.Firstly,a visual road sign based on arc angle information coding is designed.The road sign structure is simple and easy to identify.The coding method adopts Hamming coding,which has proper coding capacity,and the coding capacity has scalability.At the same time,the road signs have rotation invariance.It has strong robustness to a larger angle of view and a farther shooting distance.The road sign detection structure is used to detect the road signs.The road sign detection algorithm is implemented by VS2013 and Open CV and the experiments were carried out.After that,the signpost image is tilted,image filtered and processed,and segmented.The road sign is identified by detecting the arc angle inside the road sign,and the road sign recognition algorithm is realized by Matlab,and relevant recognition experiments are carried out.Secondly,the camera imaging model is analyzed,and the camera calibration is performed by Zhang's calibration method.The VS2013 and Open CV are used to realize the road camera calibration,and the camera internal parameters and distortion coefficients are obtained.Finally,aiming at the lack of distance information in monocular visual,based on the measurement model of the monocular vision positioning system,a three-dimensional positioning model is established.The relative positional relationship between the camera and the road sign is obtained by using the position information obtained by the road sign recognition and coordinate transformation.For external parameters that cannot be used by a mobile robot in a moving field of view,The S-W transformation model based on the positioning vector is established,and the conversion of the camera coordinate system to the world coordinate system is completed,thereby the indoor positioning of the mobile robot is realized.The mobile robot positioning algorithm is realized by VS2013 and Open CV,and the experimental analysis is carried out.It is verified that the monocular vision positioning system has no cumulative error and can ensure high positioning accuracy.