An Optimal Segmentation Method Using Jensen–Shannon Divergence via a Multi-Size Sliding Window Technique
Qutaibeh Katatbeh, José Martı́nez-Aroza, Juan Francisco Gómez-Lopera, D. Blanco · Entropy · 2015
In this paper we develop a new procedure for entropic image edge detection. The presented method computes the Jensen–Shannon divergence of the normalized grayscale histogram of a set of multi-sized double sliding windows over the entire image. The procedure presents a good performance in images with textures, contrast variations and noise. We illustrate our procedure in the edge detection of medical images.