Perceptual image processing for digital edge linking
Jamy J. Li, Jonathan Randall, L. Guan · 2004
This paper presents two methods of digital edge linking that have a reasonable correlation with psychological properties employed by human users. In the first method, each endpoint is linked to the point judged to be its best match under the criteria of alignment and proximity. The second method uses two edge images that have undergone hysteresis: a " high image" which uses high thresholds, and a "low image" which uses low thresholds. The idea is to use the high image as a guide for deciding which edges from the low image to express in the output image. Both methods achieve moderate improvements in restoring "gappy" edges in the sample images presented.