A new component based algorithm for newspaper layout analysis
Fei Liu, Yupin Luo, Motoki YOSHIKAWA, Dongcheng Hu · 2002
The aim of the layout analysis is to extract the geometric structure from a document image. It is a progress of labeling homogenous regions of a document image. In order to present a complex newspaper layout analysis, this paper proposes a new component based bottom-up algorithm. With a novel homogeneity related definition of distance, it maintains a dynamic minimal distance mechanism to decide the components merging sequence. Under the restricting rules generated from the newspaper layout heuristically, we derive the preferred analysis result. Experimental results reveal the proposed approach is effective.