Algorithm A1 Normal representative keyset (NRK)

Input: The set of normal WSIs, $H^{\mathrm{Normal}}=\{H_1,H_2,\dots ,H_N\}$, and a similarity threshold $\sigma \in \left(0,1\right)$.

Step 1: for $i=1,2,\dots ,N$ do

Set $NR{K}_i=\mathrm{D}\mathrm{F}\mathrm{I}\left(H_i\right)$, where the module DFI is given in Algorithm A2.

End (for)

Step 2: Set $NRK\hspace{0.25em}=\mathrm{D}\mathrm{F}\mathrm{I}\left({\displaystyle \underset{i=1}{\bigcup ^N}}NR{K}_i\right)$= {$k_1,k_2,.,k_m$}.

Output: $NRK\hspace{0.25em}=$ {$k_1,k_2,.,k_m$}.