Human perception of trademark images: Implications for retrieval system design

A crucial aspect of shape similarity estimation is the identification of pe rceptually significant image elements. in order to understand more about th e process of human segmentation of abstract images, a sample of 63 trademar k images was shown to several groups of students in two experiments. Studen ts were first presented with printed versions of a number of abstract trade mark images, and invited to sketch their preferred segmentation of each ima ge. A second group of students was then shown each image, plus its set of a lternative segmentations, and invited to rank each alternative in order of preference. Our results suggest that most participants used a relatively sm all number of segmentation strategies, reflecting well-known psychological principles. Agreement between human image segmentations and those generated by the ARTISAN trademark retrieval system was quite limited; the most comm on causes of discrepancy were failure to handle texture and incorrect group ing of components into regions. Ways of improving ARTISAN's ability to mode l human segmentation behavior are discussed. (C) 2000 SPIE and IS&T. [S1017 -9909(00)00104-5].