The precise multimode GNSS positioning for UAV and its application in large scale photogrammetry

In this paper,we use multimode GNSS instead of single GPS constellation to resolve the three exterior line elements of each image.The principles of differential GNSS positioning and GNSSsupported aerial triangulation(AT)are presented and an implementation case is demonstrated.With multi-constellation system,the number of visible satellites is significantly increased and the geometry distribution of the satellites is well improved.The positioning accuracy and robustness are therefore getting better compared to GPS positioning.Experimental results show that differential GNSS has remarkable increment on the integer rate of ambiguity solution when GPS has few number and low elevation angle of satellites.The combined AT adjustment of GNSS resolution and 10 ground control points(GCPs)achieve the horizontal accuracy of±18 cm and vertical accuracy of±23 cm for the check points,which are comparable with traditional bundle adjustment with dense GCPs and better than GPS-supported AT.The achieved accuracy also satisfies the requirement for 1:500 topographic maps with the bonus of 84%GCPs eliminated.In conclusion,GNSS-supported AT is of feasibility and superiority for large scale Unmanned Aerial Vehicle-based photogrammetry.

VNLSTM-PoseNet: A novel deep ConvNet for real-time 6-DOF camera relocalization in urban streets

Image-based relocalization is a renewed interest in outdoor environments,because it is an important problem with many applications.PoseNet introduces Convolutional Neural Network(CNN)for the first time to realize the real-time camera pose solution based on a single image.In order to solve the problem of precision and robustness of PoseNet and its improved algorithms in complex environment,this paper proposes and implements a new visual relocation method based on deep convolutional neural networks(VNLSTM-PoseNet).Firstly,this method directly resizes the input image without cropping to increase the receptive field of the training image.Then,the image and the corresponding pose labels are put into the improved Long Short-Term Memory based(LSTM-based)PoseNet network for training and the network is optimized by the Nadam optimizer.Finally,the trained network is used for image localization to obtain the camera pose.Experimental results on outdoor public datasets show our VNLSTM-PoseNet can lead to drastic improvements in relocalization performance compared to existing state-of-theart CNN-based methods.