Yes, you’ll use empty game objects for all sorts of purposes, including as “containers” of sorts. The container or “higher” up in the hierarchy is called the parent, the ones below is the child(s). Each game object has a component on top of its list in the inspector (the details pane), that is called the transform, position, rotation and scale. The values there depend on its relation to its parent. That means, 0.5 on X means that the child is offset by 0.5 to the right. Or 0.5 scale on all three vectors means that the content is half the size. Such relative-to-its-parent values in transform are called “local” because when you change its parents, they will reflect the relative position to the new parent. Game Objects that have no parent (or the scene itself as the only parent) show their “absolute” values, which are called “world space” rather than “local space”. I recommend you doodle around with a bunch of game objects, change their transforms and parent them to see what happens.
This means for you. When your image has a parent, and are scaled, it will treat its 0/0/0 transform (i.e. it aligns with its parent) as its pivot point and also the point it scales towards. Now when you move the image to some other point being the centre, it will scale towards that point. Hence when both images have the same setup (parent-child) they should scale the same way towards each their zero point. To do this, find the gizmo of the parent and set an icon, so you see where that centre is.