There’s been some small controversy lately about whether or not image search results should be displayed with data automatically attached, or only when the image is rolled over.
I’m not satisfied by either approach, since you still have to scan each image sequentially across and down the page to find anything vaguely resembling what you may be looking for -unless you are looking for e.g. ‘anything yellow,’ in which case you should perhaps have your internet licence revoked.
People forget that they aren’t able to perceive the entire visual field simultaneously at high resolution, because they usually have the freedom to make eye and head movements of the type that conventional image search demands -think about scanning a team photograph for the face of your favourite nephew: you can only do it by a serial scan process, not all at once.
Today’s invention makes use of this fact about perception, together with another one. We can identify images easily in around 150 ms. I therefore suggest that image search results be displayed one at a time, every 200 ms -as in a slide show. When you spot an image that is of interest, simply click on it to stop the process for a more detailed look.
This would enable reliable inspection of 300 images per minute: much faster than can be achieved when including the additional time taken to scan between them arrayed on a page. It would also allow images to appear at greater resolution than in their present, postage-stamp size.
(In fact, the rate of inspection could probably be increased further, without actually losing much recognition capability).