Box-Counting Dimension Analysis of Edge Images Detected with Classical Edge Detector

This paper proposes that box-counting dimension method be adopted for the analysis of edge images detected with classical edge detectors. By using this method, the authors find that the number of ln(N(l)) and ln(l) of binary image is greater than that of edge images; and at the resolution of 512×512 in pixel, ln(N(l)) and ln(l) of the edges images is more disorder than that of binary images when ln(l) < -1.5, and they all close up to a line whose slope is the value of FD when ln(l) > -1.5, and this scene appears when ln(l) < -1.2 and ln(l) > -1.2; and that edge images detected by different detectors have dissimilar fractional dimension value.