Tk makes it easy for you to do that, because it is an event-loop system. So you can start a timer, which checks for a shared variable every 10 ms (or whatever update time you need). You didn't show your code, but you probably want to reuse your threads, so you don't get a memory gain.

So the basic idea is to make your threads, before you invoke any Tk code, see PerlTk on a thread... for other precautions. Then setup your Tk gui. Then use shared variables to signal back and forth between the Tk in the main thread and the worker thread(s). To detect when a thread is done processing an image, have a shared var called for instance "$done", and have a timer in the main Tk thread, check $done every 10 ms. When $done == 1; transfer the image thru another shared var into the Tk main thread. Since Tk likes it's images to be base64encoded anyways, have the thread base64encode the image, before it puts it into the output shared var. See also Tk-with-worker-threads for an example of making reusable threads that sleep when not used. This technique prevents any increase in memory caused by continually creating and destroying threads.( You cannot create/destroy threads with Tk anyways, since Tk is not thread-safe, and you need to create ALL threads before any Tk code is invoked in the main thread).