Fast extraction of scene structure based on gradient runs analysis

The paper presents a novel edge detector for fast localization of image features and homogeneous regions extraction. A new approach based on gradient runs analysis is used to localize edges. Runs are sequences of pixels having the same property, such as equal intensity, adjacent in a row or column. The segments of a piecewise-linear approximation of a row/column of a grayscale image are named gradient runs. The gradient runs representation speeds up and facilitates image analysis. The runs are linked into a planar graph reorganizing the edge information for topology analysis. The algorithm of edge detection using gradient runs and an algorithm sketch of edge graph construction based on appropriate gradient runs linkage are presented.