Image segmentation and linear feature identification using rectangular block decomposition
Stavros J. Perantonis, Basilis Gatos, N. Papamarkos · 2002
The decomposition and segmentation of binary images using rectangular blocks of foreground pixels as primitives is discussed and two algorithms related to these tasks are proposed. A novel fast method for evaluating the Hough transform by taking advantage of these primitives is introduced, which can be used to accelerate detection of prevalent linear formations in binary images. The resulting acceleration is demonstrated in two image processing experiments related to skew detection of digitized documents and object axis identification.