Using a CRC hash for the files (stripping sharepoint meta data) seems appropriate. As long as the CRC is long enough to provide a statistically unique hash based on the number of files you are indexing.
Why are you trying to combine them for a folder based CRC? How are you planning on combining them? If you are thinking of just summing the CRCs for the folder value I'd say that wouldn't be guaranteed to be unique.
I think it would be better to create a dictionary using the CRC as the key and then simply use Dictionary.ContainsKey for look up/comparison
or watch for duplicate entries via the Dictionary.Add catching ArgumentExcptions (key already exists)