Iv Sjsn added subsection_Video_Graph_Data_Structure__.tex  almost 9 years ago

Commit id: 765ec0ca125cc64ec325c6aa13b2c5028ba5ce8a

deletions | additions      

         

\subsection{Video Graph Data Structure}  A video sequence is partitioned into a set of non-overlapping supervoxel regions $S = \{s_0,s_1,..,s_N\}$.   The supervoxel is mapped into graph: $f_{map}: S \mapto G(V,E)$, where each node $e_i$ has following attributes:  \begin{itemize}  \item $size_i$ : the total number of cells making up the region $s_i$, $\mathbb{R}$.  \item $min_i$ : the minimum most location of 3D bounding box point in cell-coordinate $\mathbb{R}^3$.  \item $max_i$ : the minimum most location of 3D bounding box point in cell-coordinate $\mathbb{R}^3$.  \item $mean_i$: the mean cell location, $\mathbb{R}^3$.  \item $hist_i$: the bow histogram over the local features $f_i \in s_i$, $\mathbb{R}^{4k} \ntimes 5$  \item $sparsity_i$: the number of non-zeros bins in the bow histogram $hist_i$.   \item $$  \end{itemize}