The really magical part of the brain is its ability to selectively throw away massive amounts of incoming data before it even begins processing it. We develop a system by which to determine what portions of incoming data are worth considering, and the rest goes into really shoddy temporary storage for a few seconds (just in case) before being destroyed forever. We shouldn't try to get computers to process as much information as we pretend we do, we should train them to recognize what can bear to be ignored based on an overall feeling of the incoming sample.
Imagine looking at a really pixelly jpeg that has a flesh colored blob in it with two darker blobs at the center and a darker area circling around the top. Based on the hundreds of thousands of times you've seen that type of information, I'll bet you could very effectively decide which portion you'd want to increase clarity on. I'll bet we could get a computer to decide, too. And maybe you're wrong about it being a face, and it's actually a satellite photo of Hawaii in inverted colors. Then all you get is the bad stuff in temp storage or a chance to re-scan.