About point cloud position retrieval

[lkf263]

I appreciate your great work. I am curious about how you achieved the visualization of point cloud position retrieval, similar to Figure 4 or5. If you could provide insights into this, I would be immensely grateful.Thank you!

[jac99]

Hi,

I've used the Jupyer notebook in the attached file to generate point cloud plots (separate plot for each point cloud - based on the kNN search results saved to a .txt file). Then, I've combined separate plots in Latex: point_cloud_visualiazation_code.zip

\captionsetup[subfigure]{labelformat=empty}

\begin{figure*}[htbp] \centering \setcounter{subfigure}{0} \subfloat[query]{% \includegraphics[width=2.7cm,height=2cm,trim={1.5cm 1cm 0.5cm 2cm},clip,cfbox=black 0.5pt 0.5pt]{images/nn_1.png}} \hfill \subfloat[dist=1.13 TP]{% \includegraphics[width=2.7cm,height=2cm,trim={1.5cm 1cm 0.5cm 2cm},clip,cfbox=mygreen 0.5pt 0.5pt]{images/nn_3.png}} \hfill \subfloat[dist=1.40 TP]{% \includegraphics[width=2.7cm,height=2cm,trim={1.5cm 1cm 0.5cm 2cm},clip,cfbox=mygreen 0.5pt 0.5pt]{images/nn_5.png}} \hfill \subfloat[dist=1.44 TP]{% \includegraphics[width=2.7cm,height=2cm,trim={1.5cm 1cm 0.5cm 2cm},clip,cfbox=mygreen 0.5pt 0.5pt]{images/nn_7.png}} \hfill \subfloat[dist=1.45 TP]{% \includegraphics[width=2.7cm,height=2cm,trim={1.5cm 1cm 0.5cm 2cm},clip,cfbox=mygreen 0.5pt 0.5pt]{images/nn_9.png}} \hfill \subfloat[dist=1.82 FP]{% \includegraphics[width=2.7cm,height=2cm,trim={1.5cm 1cm 0.5cm 2cm},clip,cfbox=red 0.5pt 0.5pt]{images/nn_11.png}}

\caption{Nearest neighbours search results in Oxford evaluation subset. The leftmost column shows a query point cloud. Other columns show its five nearest neighbours. \emph{dist} is an Euclidean distance in the descriptor space. TP indicates true positive and FP false positive.} \label{fig:search_results} \end{figure*}

\captionsetup[subfigure]{labelformat=parens}

[jac99]

Closing