Abstract：A predictive erroneous matching risk minimization based feature selection method is proposed for visual SLAM (simultaneous localization and mapping).It uses predictive erroneous matching risk to measure the influence of newly detected features on the oncoming feature matching process.Then,based on a multi-layer ranking method,the new features with lower erroneous matching risk and higher repeatability are selected for initialization with priority.This method can adaptively select good features that are not prone to be erroneously matched according to the uncertainty in state estimation. Therefore,the convergency and consistency of the SLAM algorithm can be ensured.The comparative experiment results on a mono-SLAM system validate that the proposed method has significant advantages over the existing methods in reducing erroneous matching rate and ensuring the correctness of state estimation.
[1] Se S,Lowe D,Little J.Mobile robot localization and mapping with uncertainty using scale-invariant visual landmarks[J]. International Journal of Robotics Research,2002,21(8):735-758.
[2] Davison A,Reid I D,Molton N,et al.MonoSLAM:Real-time single camera SLAM[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2007,29(6):1052-1067.
[3] Harris C,Stephens M.A combined corner and edge detector[C]//4th Alvey Vision Conference.Manchester,UK:1988:147-151.
[4] Lowe D G.Distincdve image features from scale-invariant keypoints[J]. International Journal of Computer Vision,2004,60(2):91-110.
[5] Clemente L A,Davison A J,Reid I D,et al.Mapping large loops with a single hand-held camera[C]//Robotics:Science and Systems Conference.Atlanta,GA,USA:MIT Press,2007:1-8.
[6] Smith P,Reid I D,Davison A J.Real-time monocular SLAM with straight lines[C]//17th British Machine Vision Conference.Edinburgh.UK:2006:17-26.
[7] Molton N,Reid I D,Davison A J.Locally planar patch features for real-time structure from motion[C]//15th British Machine Vision Conference.Kingston,UK:2004:1-14.
[8] Chekhlov D,Pupilli M,Mayol-Cuevas W,et al.Real-time and robust monocular SLAM using predictive multi-resolution descriptors[M]//Lecture Notes in Computer Science (vol.4292).Berlin,Germany:Springer-Verlag,2006:276-285.
[9] Civera J,Davison A,Montiel M.Inverse depth parametrization for monocular SLAM[J]. IEEE Transactions on Robotics,2008,24(5):932-945.
[10] Lemaire T,Berger C,Jung I-K,et al.Vision-based SLAM:Stereo and monocular approaches[J]. International Journal of Computer Vision,2007,74(3):343-364.
[11] Davison A J.Active search for real-time vision[C]//IEEE International Conference on Computer Vision.Piscataway,NJ,USA:IEEE,2005:66-73.
[12] Chli M,Davison A J.Active matching for visual tracking[J]. Robotics and Autonomous Systems,2009,57(12):1173-1187.
[13] Mikolajczyk K,Tuytelaars T,Schmid C,et al.A comparison of affine region detectors[J]. International Journal of Computer Vision,2005,65(1/2):43-72.
[14] Chekhlov D,Pupilli M,Mayol W,et al.Robust real-time visual SLAM using scale prediction and exemplar based feature description[C]//IEEE Conference on Computer Vision and Pattern Recognition.Piscataway,NJ,USA:IEEE,2007:430-436.
