That's a great suggestion, @swicher1. I did some testing and see how a student using the Canvas Student app to submit a file to an assignment via the camera generates an image file even if the item is all text.
A solution would be for the DocViewer to automatically apply optical character recognition to all image and .pdf files that it handles. I don't know how feasible that is but I do know how to begin that conversation: Use the ideas and themes process:
How do Ideas and Themes work in the Instructure Community?
Creating an idea does not provide an immediate solution to the problem you described, though it is a solution. I’m going to mark it as such and hope that our colleagues who have a similar challenge will find the best way to contribute. If you decide to create an idea, please link it here so mode community members can support it.