New estimation method of the contrast parameter for the Perona–Malik diffusion equation
M. Borroto-Fernández, Manuel González-Hidalgo, Ángela M. León‐Mecías · Computer Methods in Biomechanics and Biomedical Engineering Imaging & Visualization · 2014
The aim of this contribution is to make an efficient smoothing algorithm that preserves edges and provides valuable information for any segmentation process. The non-linear anisotropic diffusion (AD) model of Perona–Malik is considered to enhance the edges in the process of diffusion through a variable diffusion coefficient. However, the diffusion coefficient is very sensitive to the so-called contrast or gradient threshold parameter. This article proposes a novel methodology for the estimation of this contrast parameter based on a partition of the image using the K-means algorithm and a least-square fit to approximate the diffusion coefficient (KMLS). The experimental results show that the quality of the edge detection process improves when the proposed algorithm in the smoothing AD is used instead of the traditional techniques. The comparison is performed using two objective edge detection performance measures, the so-called Pratt's figure of merit and the symmetric average distance. Both measures show great improvements if we use the Perona–Malik equation with the KMLS estimator.