You could copy those several images to a bigger one, specifying
appropriated areas, or regions.
But you have higui. Nowadays, OpenCV uses Qt in its "higui" basic
interface. I said "basic", because that is what it was intended to, basic
visualization for testing purposes (AFAIK).
Extending the basic higui interface is not so complicated, look at its
source code, it might help a lot.
If you need more sophisticated interfaces, try some of the projects mixing
Qt and OpenCV or even other GUI libraries.
A good hint is what guys are doing, kind of mixing the inner image
representation of both Qt and OpenCV , although those libraries have
different types for what they call an image: there are a few OpenCV raw
image types that match also raw Qt image types, so making the "data"
pointer of a QImage point to the same place as the "data" pointer of a
cv::Mat , will avoid processing resources being used on transforming from
one type (the image you will be working on using OpenCV) to the other
(QImage from Qt, or whatever you choose) for GUI representation. Those
images are CV_8UC3 and CV_8UC1.