SEO by the Sea has a post on a Microsoft patent filing for the factors that count in the ranking of images for image search results. The patent filing and the post make a good reading for those who want a good idea on the kind of factors that search engines take into consideration while ranking images.

I would not agree with many of the relationships that the patent seeks to establish between the images and the text that surrounds the images. For one reason, relying too much on the link information and the meta data is only a technological workaround for our inability to compute image relevance strictly based on the image itself. To this extent a recent patent filing from Google on using a VisualRank to computer links between similar images and rank them makes a lot more sense.

Nonetheless, the factors mentioned here give great deal of information on what factors are used by image search crawlers to rank images and can be put to great use.