Analysis of fragmented images perception: local features and global description

Abstract

Analysis of experimental investigations of the perception of incomplete images is presented. It illustrates two different approaches to work of the brain mechanisms involved: one approach is based on the perception of whole images and another on local informative features. These approaches describe two different mechanisms, which are possibly used by brain systems for incomplete image recognition. Performance on the Gollin test (measuring recognition thresholds for fragmented line drawings of everyday objects and animals) depends upon recognition based on image informational-statistical characteristics. We suggest that recognition thresholds for Gollin stimuli in part reflect the extraction of signal from noise. The brain uses local informative features as an additional source of information about them. We have suggested that fragmented images in the Gollin-test are perceived as whole structures. This structure is compared with a template in memory which is extracted with the help of selective attention mechanism in accordance with a matched filtration model. The Gollin-test is a tool for differential diagnosis of a various forms of cognitive disorders.