Cite as:
Gurciniene O, Vanagas V, Soliuñas A, 1998, "Modelling of the effect of pattern symmetry, similarity, and complexity on its recognition time" Perception 27 ECVP Abstract Supplement
Modelling of the effect of pattern symmetry, similarity, and complexity on its recognition time
O Gurciniene, V Vanagas, A Soliuñas
At the 14th, 16th, and 17th ECVP (1991 Perception 20 Supplement, 124; 1993 Perception 22 Supplement, 50 - 51; 1994 Perception 23 Supplement, 44), we reported on the influence of pattern (figure composed of horizontal and vertical line segments) symmetry, similarity, and complexity on its recognition probability in psychophysical tachistoscopic experiments on human subjects. To determine the mechanism underlying this influence, a model of an active recognition system of such simple formalised patterns is introduced. Recognition in the model is regarded as a process of creation and verification of hypotheses by matching the features (line segments) of presented and hypothetical patterns. If the line segment in the presented pattern is absent, all hypothetical patterns that possess this line segment are rejected and vice versa. The recognition time is defined by the number of verified hypotheses and that depends on the efficiency of features used for pattern description. The most effective features are those which allow a reduction in the number of hypotheses by one half. A computer simulation was carried out with the same patterns as those used in the psychophysical experiments. In most cases the obtained results correlated with the psychophysical data. This allows us to propose that similarity, symmetry, and complexity define the efficiency of features used for pattern description in the human visual system, and how these influence recognition time.
These web-based abstracts are provided for ease of seaching and access, but certain aspects (such as as mathematics) may not appear in their optimum form. For the final published version of this abstract, please see
ECVP 1998 Abstract Supplement (complete) size: 2107 Kb
