Dataset for N.Radwell et. al. "Hybrid 3D ranging and velocity tracking system combining multi-view cameras and simple LiDAR", Scientific Reports (2019).

This data set is separated by Figure number, such that each folder contains the data relating to each file within the manuscript. Folder by Folder:

Figure 2:

The input data are the .jpg files. VisualSFM (http://ccwu.me/vsfm/) is used to produce the .sift files which contain the feature definitions and the .mat files which are the matches between features. The reconstruction based on these is a .nvm file which contain the reconstructed camera positions as well as the 3D point cloud. The point cloud is also provided in .ply format. A simple python plot of the data is also provided (NVM_3D_Plotter.py). For further information on any of the files provided, see the documentation on the VisualSFM website.

The images in sections II, III and IV are also provided, and these have been generated by the python code 'Multiview_Figure.py'.

Figure 3:

Figure 3a is the same as in Figure 2. Parts d) and e) are generated using a python script (LiDAR_Data.py) with LiDAR timing data (output2.csv). Figures 3b) c) and f) are generated using a python script (Data_Fusion_Algorithm_General.py) together with a point cloud (Points.nvm) and LiDAR timing data (output.csv). We note here that the exact location of the peaks from the nvm file are not matching those in the paper, this is probably due to the non-deterministic point cloud reconstruction from VisualSFM - at some point we may have recalculated the point cloud, causing it to overwrite the old data, and due to the random seed in the reconstruction, returns a slightly different (though quantatively similar) histogram of depth.

Figure 4:

THe input data are the .jpg files, from which VisualSFM produces a .nvm point cloud. A python script (Data_Fusion_Algorithm_Cycling.py) scales the LiDAR data (LiDAR_CSV.csv) and the NVM file (Point_Cloud.nvm) to produce a 'fused' point cloud (pointCloudScaled.csv) as well as the camera pixel aligned version (camPointsScaled.csv). Figures 4d and 4e are generated by a python script (Cycling_CSV_Plot.py) which also saves the peak position and speeds(PeakCentroidPositions.csv).

This data is then conerted to the composite image (Composite_Imge_new.jpg) using a LabVIEW program (Composite_Image_Creator.vi).

Figure 5:

The data is treated in the same way as in figure 4, with 'Walking' and 'Stationary' related filenames.

Flatness Calculations:

The poster board point clouds have been isolated and saved to .nvm files which are in the 'PosterFlatness' folder. These are analysed in the python file 'Poster_Board_Flatness.py', as described in the manuscript.