engleski

Line simplification based on the coefficient of linear correlation

The goal of simplifying lines in digital cartography is to reduce the information content, whereby the simplified line should be similar to the original, i.e. it should leave a similar visual impression on the observer as the original, and metric deviations should remain within the predetermined limit. Experiences from the theory of pattern recognition, but also from the border areas of geodesy and geophysics, indicate a strong connection between the similarity of visual impressions and the similarity of shapes. If one abandons the usual limitation in statistics that the linear correlation coefficient is applied only to stochastic quantities and applies it to quantities that define geometric configurations (eg coordinates), a new interpretation of that coefficient is reached. Thus, it becomes a measure of shape similarity. This paper presents the principles of simplification of individual lines based on the maximum correlation between the original and simplified lines. The possibility of applying the procedure is illustrated by examples. At the same time, it was shown that this method retains the characteristic shape of the line much better than methods based only on metric criteria.

cartographic generalization ; line simplification ; coefficient of linear correlation ; similarity of forms

nije evidentirano