First, have you considered using ancestry.com for your census work? Ancestry makes all those decisions for you, sets up the facts, enters the source and attaches the citation to each person, attaches the media - and names them and captions them - all automatically - you don't have to do a thing. I can tell you from years of doing census work the "old-fashioned" way - via micro-film readers - the new ancestry way is a god send. (I'm not a salesman, just a satisfied customer - with a few gripes here and there - like the citations are too wordy)
At any rate, whatever kind of a systematic way that works for you is fine. There is no "right" or "wrong" way.
FTM used to assign file names for census media "1830 United States Federal Census- 001.jpg for the first file you downloaded from that census. Then each file you added after that was 002, 003, etc. Now ancestry uses file names like "1870 United States Federal Census - John Smith.jpg" Both names drive me crazy if I were to use them as I see no need to spell out United States or even use the word Federal.
Ancestry gives a caption for the person or head of household something like "John Smith - 1830 United States Federal Census" - for the person or head of household.
Personally, I could care less what ancestry uses. Since I always access an image attached via a citation from the Person / Fact page or the Source / Citation workspace, I never use media file names or captions for citation media. They could be random numbers as far as I'm concerned. We just had a conversation on this board about persons with quite different views. Other people go to great lengths to use file names, and I presume captions, according to a system of their own devise.