We are using a template descriptor on the image in order to try and find the object. However, we
have a sparse 3D point clouds of the world that is not used at all when looking for the object in the images.
Considering there are many false alarms during the detection, we are interested in exploring how to combine the
detections on the image with the 3D point clouds in order to reject some detection outliers. In this experiment we
use semi-direct-monocular visual odometry (SVO) to provide 3D points coordinates and camera poses to project
3D points to 2D image coordinates. By un-projecting points in the tracking on the selection tree (TST) detection
box back to 3D space, we can use 3D Gaussian ellipsoid fitting to determine object scales. By ruling out different
scales of detected objects, we can reject most of the detection outliers of the object.
GUO Li (郭立)
. A Comprehensive Method to Reject Detection Outliers by Combining Template Descriptor with Sparse 3D Point Clouds[J]. Journal of Shanghai Jiaotong University(Science), 2017
, 22(2)
: 188
-192
.
DOI: 10.1007/s12204-017-1820-x
[1] LEE T, SOATTO S. Learning and matching multiscaletemplate descriptors for real-time detection, localizationand tracking [C]//In Proceedings of the IEEEConference on Computer Vision and Pattern Recognition(CVPR). Colorado, USA: Institute of Electricaland Electronic Engineers, 2011: 1457-1464.
[2] BABENKO B, YANG M H, BELONGIE S. Visualtracking with online multiple instance learning [C]//InProceedings of the IEEE Conference on Computer Visionand Pattern Recognition (CVPR). San Fransico,USA: Institute of Electrical and Electronic Engineers,2009: 983-990.
[3] HINTERSTOISSER S, LEPETIE V, ILIC S, et al.Dominant orientation templates for real-time detectionof texture-less objects [C]//In Proceedings of the IEEEConference on Computer Vision and Pattern Recognition(CVPR). Miami, USA: Institute of Electrical andElectronic Engineers, 2010: 2057-2069.
[4] LEE T, SOATTO S. TST/BTD: An end-to-end visualrecognition system [R]. Los Angeles: UCLA TechnicalReport, 2010.
[5] BL¨oSCH M, WEISS S, SCARAMUZZA D, et al.Vision based MAV navigation in unknown andunstructured environments [C]//Proceeding IEEEInternational Conference on Robotics and Automation.Alaska, USA: [s.n.], 2010: 21-28.
[6] WEISEE S, ACHTELIK M W, LYNEN S, et al.Monocular vision for long-term micro aerial vehiclestate estimation: A Compendium [J]. Journal of FieldRobotics, 2013, 30(5): 803-831.
[7] SCARAMYZZA D, ACHTELIK M, DOITSIDIS L, etal. Vision-controlled micro flying robots: From systemdesign to autonomous navigation and mapping inGPS-denied environments [J]. IEEE Robotics and AutomationMagazine, 2014, 21(3): 26-40.
[8] FORSTER C, LYNEN S, KNEIP L, et al. Collaborativemonocular SLAM with multiple micro aerial vehicles[J]. Proceeding IEEE/RSJ International Conferenceon Intelligent Robots and Systems, 2013, 8215(2):3962-3970.
[9] FORSTER C, PIZZOLI M, SCARAMUZZA D.Air-ground localization and map augmentation usingmonocular dense reconstruction [C]//ProceedingIEEE/RJ International Conference on IntelligentRobots and Systems. Tokyo, Japan: IEEE, 2013: 592-625.
[10] FORSTER C, PIZZOLI M, SCARAMUZZA D. SVO:fast semi-direct monocular visual odometry [C]//IEEEInternational Conference on Robotics and Automation(ICRA). Hong Kong, China: IEEE, 2014: 624-675.
[11] DEVERNAY F, FAUGERAS O. Straight lines haveto be straight: Automatic calibration and removal ofdistortion from scenes of structured environments [J].Machine Vision and Applications, 2001, 13(1): 14-24.
