Is that correct or there is a more subtle answer?
We just have to check it against the multiplication rule of the two groups, to see whether it's a homomorphism. The product $(h_1,k_1)\cdot (h_2,k_2)$ in the first group is $(h_1\varphi(k_1)(h_2),k_1k_2)$. In the second group, we have the product of $(h_1,\sigma(k_1))$ and $(h_2,\sigma(k_2))$, which is
$$f(h_1,k_1)\cdot f(h_2,k_2) = (h_1,\sigma(k_1))\cdot (h_2,\sigma(k_2)) = (h_1\varphi'(\sigma(k_1))(h_2),\sigma(k_1)\sigma(k_2))$$
$$(h_1\varphi'(\sigma(k_1))(h_2),\sigma(k_1)\sigma(k_2)) \stackrel{?}{=} (h_1\varphi(k_1)(h_2),\sigma(k_1k_2)) = f((h_1,k_1)\cdot (h_2,k_2))$$
That works in the second coordinate, but not in the first; we would need $\varphi'\circ\sigma=\varphi$, and we instead have $\varphi'=\varphi\circ\sigma$. Indeed, this isn't correct.
How do we repair it? Simple - use $f(h,k)=(h,\sigma^{-1}(k))$ instead. Or, equivalently, turn the arrow around.