This is a great suggestion. For any topics like these, starting in very low dimensional space and working through an example is a fantastic way to understand it.

I'd also suggest not only trying to imagine this in your head: implement it in code. That's how I truly learn these types of things.

For extra fun, you can make it into a game. After you've implemented the filter, create a way for a human to input a guess at the filtered output. Score it depending on how close it is. This way you get to practise your intuition for it too.