What do users see? Exploring the cognitive nature of functional image retrieval

Images cannot just be labeled. The vexing issues surrounding functional access to digitized images have come to the fore as a major concern in the realm of information retrieval. Constant overlap as well as a lack of consistent membership among and between images continues to challenge the efficacy of retrieval systems development. Analysis of user viewing and categorization furnishes additional evidence that what viewers see is not necessarily the objects pictured, may not even be present in an image, or represents a wide variety of possible descriptions. This investigation looks at how users view images in relation to how they can be described (categorized) and in what manner they match other images in the same collection. The results provide an indication that what viewers see depends as much on who they are as it does on what they see. While traditional methods of image retrieval have mostly focused on indexing techniques that encompass object descriptions and more recent approaches through contentbased aspects of digitized imagery, it appears that these systems could be improved by enabling subjective engagement with retrieved images to identify additional access points derived from the adumbrative, impressionistic, and many times abstract nature of user cognitive engagements with the images retrieved.