Research on Improved Visual SLAM Localization Based on Feature Points
In order to improve the low localization accuracy of simultaneous localization and mapping(SLAM)system in low tex-ture environment,this paper proposes an improved oriented FAST and rotated BRIEF(ORB)feature point extraction strategy and key frame selection mechanism.Firstly,multi-scale analysis and feature detection method based on local gray level are used to over-come the shortcomings of lack scale and rotation description in general ORB algorithm.Secondly,an image information enhancement method based on Gaussian blur is proposed to solve the problem that the texture information of the traditional ORB feature point ex-traction method is easy to fail in special environment,and the image is segmented to evenly distribute the feature points.Finally,in order to eliminate the inferior key frames,a key frame selection mechanism combining time factor and feature point number factor is designed.The proposed method is transplanted to the ORB_SLAM2,and testing on the TUM dataset,the experimental results show that the localization error of the visual SLAM system is reduced by 14.688％on average,which proves the effectiveness of the proposed method.
万方数据
维普
