A theory of parameter-dependent connected components of gray images and segmentation

Abstract

A new definition of the connected components of gray images which depends on two parameters /spl epsiv/ and /spl delta/ is introduced. The definition takes into account both the gray values of the pixels and the differences of the gray values of the neighboring pixels. An algorithm to partition an image into its (/spl epsiv/,/spl delta/)-components and several experimental results are given. A practical advantage of the definition is that a segmentation of an image can be done by finding its (/spl epsiv/,/spl delta/)-components for some appropriate values of /spl epsiv/, /spl delta/. Also, we may study an image globally through the interrelationships of the (/spl epsiv/,/spl delta/)-components rather than that of the pixels. Thus our theory provides a possible method of transition from low level computer vision to an intermediate level vision.