Search engines are largely unable to pick an original from its clones.
Search engines routinely discount "duplicate content" from their result pages, in order to provide a more varied user experience. Image recognition technology allow search engines to identify duplicate images as well, and purge them out of their Image searches.
How do search engines distinguish the original content from its duplicates? The short answer is: "THEY DON'T." They are not able to recognize an original form a copy. The image that beats the clones is the one with the greatest "weight" as measured by imponderables like backlinks, anchor text, and page rank - the exact kind of scheme that repeated repinning of an image will favor. One could predict that a much-repinned image on Pinterest will handily supplant that of the original, under current search engine practices.
Oftentimes, a page with a link to an image on a third-party website will rank higher than the page with the actual image. It's safe to say that the process is quite imperfect.
We don't know how search engines will handle images hosted on copyright-infringement platforms like Pinterest in their ranking algorithms. It is certain, however, that Pinterest's copies of one's images represent a clear danger to the ranking of one's original images. They almost certainly will, in many instances, supplant the original in terms of ranking and hurt its visibility and credibility on the internet, unless a search engine manually punishes Pinterest-hosted images with a penalty.
When a artist decides whether to allow Pinterest users to scrape their images and upload them to Pinterest's servers, he or she will often be blinded by flattery, and noble notions of free sharing. Few will consider the possibility that their original image's rankings will be hijacked by Pinterest's many copies, diverting their hard-earned traffic to Pinterest's benefit.