Imagine you have 50 identical pennies on a table top.
An uncompressed file describes each penny in detail.
A JPEG describes one penny in detail, and then tells you where the other 49 are located on the table.
Again, I'm sorry, but JPEG doesn't work like that either. Fractal compression algorithms are conceptually similar to the method you describe, but there aren't any popular implementations of those.
I could attempt a description of how JPEG "really" works if people are interested?
S
