Visual processing of configuration-dependent spatial characteristics of shapes and patterns. A model useful in the study of the role of the departure from circularity or dispersion of shapes in human visual perception.

In this work a theoretical model was used in combination with testings on normal subjects to get more insight in the role of the departure from circularity or dispersion of the shapes in visual perception. The model was inspired by the observation that the intensity of the effect of a given level of contrast of a shape usually increases, for the same area, with the shape being better concentrated around a center. The model introduces as a measurable characteristic the degree of concentration or dispersion of a shape with respect to a center. The measure was based on the maximum of the convolution integral of the characteristic function of the shape with the weighting function 1/2 pi r, r being the distance between the point of convolution and the surface element to be integrated. A program for the calculation of the degree of concentration of figures and other related processing operations was developed in Turbo Pascal language on a 486 PC. The program included the possibility to generate various figures and to operate on them various transformations such as strangulation, fragmentation with separation of fragments. The model introduces a center of the figure, the point best surrounded by the whole figure, with a geometric and visual significance, as resulting from the good concordance between its calculated and perceived positioning in different relatively simple shapes. In symmetrical compact figures subjected to a central separation or narrowing two centres appear entering the two resulting nuclear parts; a good concordance between model and perception was again observed in this transition to two centres and their subsequent positions in the two nuclear parts. In accord to model prediction, testings showed a very pronounced dependence of the summation efficiency over a contrasting area on the degree of dispersion of the area. This is reflected in the drastic decrease upon figure dispersion of the intensity with which a given brightness or colour contrast is perceived. Thus, the model gives a better explanation and a more efficient way to approach the great capacity of the visual system to disclose more compact shapes or agglomeration zones in a complex visual scene. This capacity is to a large extent due to the increase in the intensity with which a given contrast is perceived, occurring in these conditions. This intensity, which strongly depends on the degree of concentration or dispersion of the figure, becomes an important additional signal leading to the accentuation of the difference between compact and rarefied shapes. The model based on the degree of concentration determined around a centre, although useful for finding a centre and applicable satisfactorily to many shapes, do not cover well all aspects of shape dispersions. In shapes without a dominant central part the confrontation model-testing showed an important involvement in global perception of all local concentrations, not only central but also peripheral, the later underestimated in our model. The model can be however improved by taking into account also such local concentrations.