The cost will need to be higher. Software is expensive to produce, and also needs to be maintained as operating systems change and get updated.
But I do think there's a lot of room for a higher end service.
One that maintained a database of images of the graded cards would be good, especially if the scans were at higher resolutions. (Plus, if it could be worked out legally the images could possibly be licensed to auction houses, saving them the effort of creating scans. Not perfect, since the case could be damaged or scratched, but a possible way to recoup some of the data storage costs)
For many cards I don't think the digital fingerprint will work all that well. In new condition a manufactured item should be identical to the one made before and after. For an older worn card it would work, but if newer cards or very high grade cards are done, there would eventually be problems.
The unslabbed certification and grading is the standard for stamps. PSE will slab them, but it just hasn't gained acceptance in that market. For a good look at how it's done check out the Philatelic Foundation.
http://www.philatelicfoundation.org/.../expertize.htm
Their service starts at $27, with a select group of less expensive items being discounted to $20. Approximately 30 day turnaround unless you want it quicker which costs more.
The most difficult things will be handling stuff that won't show in a scan, and the cards that don't easily fit a particular grade.
For instance this Konetchy. There's a tiny flake of surface missing from one corner. if it's seen as paperloss then they got the grade right. And yet I've seen cards with corners more rounded than what's been lost that graded as 60
Or, cards with factory but unusual features. This Needham was rejected because of rough but factory cuts top and bottom. I can see why they'd decline grading, most people wouldn't understand the difference in edge quality between the normally sharp blade and the worn blade and sacrificial strip that caused the rough cut, and would think it was trimmed.
Both of those would give a computer program trouble.
Size variance would also be problematic, I've had one rejected for min size vertically, but another in the same submission that was narrower by a differenc that was more than the difference in height than the rejected one and it got a grade. I have a few I like a lot, but won't send in since I know they're too short to get a grade even though they're factory cut.
I'd like to see a premium service work, and work well. And I've considered doing something as well. (More of a traditional grading but having an available detailed explanation of the grade. How many times do we see cards with an MK qualifier but can't find the mark?) But looking at it as a business, it's a massive uphill climb to even get into a solid fourth place. A registry that recognized other companies might help, but probably not much. So there's a load of upfront cost, and a need for massive ongoing advertising.
The cert for raw cards idea is the workable one. Maybe it would generate enough to support the premium service, maybe not.
Steve B