The graph $H_s$ has clique number $3$ (since $H$ has clique number $2$, and a clique in $H_s$ cannot include more than one stellating vertex). It also has chromatic number $3$ (keep the $2$-coloring of $H$ and give all the stellating vertices a third color). So we can try to prove that $H_s$ is perfect directly from the definition: we can check that if $G_s$ is an induced subgraph of $H_s$, then the clique number $\omega(G_s)$ and the chromatic number $\chi(G_s)$ are equal.
Let $G_s$ be an induced subgraph of $H_s$, with $G$ the common subgraph of $G_s$ and $H$. If $G_s$ has no vertices, then $\omega(G_s)=\chi(G_s)=0$. Otherwise, if it has no edges, then $\omega(G_s)=\chi(G_s)=1$. If $G_s$ has both vertices and edges, then $2 \le \omega(G_s) \le \chi(G_s) \le 3$, and we can try to $2$-color $G_s$ by the following strategy:
- Keep the $2$-coloring of $G$ that we know exists because $G$ is bipartite.
- For each stellating vertex, try to give it a color not present on any of its neighbors.
If this strategy succeeds, then $\omega(G_s) = \chi(G_s) = 2$. If it fails, then there is a stellating vertex $x$ which has two neighbors of opposite colors. But then those two neighbors are consecutive vertices of the face that $x$ stellated - so, together with $x$, they form a triangle. In this case, $\omega(G_s) = \chi(G_s)=3$.
It follows that the stellated graph $H_s$ is guaranteed to be perfect.