| The augmented reality of mobile terminals' development relies on the rapid development of hardware.The unmarked AR of mobile is one kind of AR technology.The feature of this technology is that there aren't other markers or GPS positioning when running this system.Only the computers do is to calculate the camera movement by the relation of the neighboring pictures,and then we can get the position in space.So there are high research value and application prospects in this society to achieve the fast and stable of mobile AR.This thesis mainly studies the related aspects of mobile augmented reality technology based on SLAM visual odometry.When the real environment texture is not rich,how to construct a three-dimensional spatial map with higher stability and faster speed and positioning is the key problem which this paper to solve.In terms of stability,the test get the camera's pose by calculating the images' line feature.The reason why we choose this method is that low texture may make point features fail,because those point features may not be detected.In order to getting the more stable we choose the Line Segment Detector line feature detection method because of stability.In order to speed up the calculation,we choose the direct method.In this paper we choose the LSD-SLAM visual odometry which is based on the direct method.Comparing with the feature point method,there is less computation in the direct method.Finally,we integrate the line features into the LSD-SLAM visual odometry.Then our experiment use TUM-RGB data set to test the algorithm.The experimental results show that adding line features to the visual odometry of direct method can not only improve the robustness and accuracy of camera pose estimation,but also accelerate the accuracy of monocular vision matching.Finally,the experiment is applied to the scene of augmented reality furniture display,and a mobile-end augmented reality furniture display system is designed to test the performance and speed of the experiment. |
PDF Full Text Request