Clusterização Baseada na \varphi-Divergência Aplicada à Segmentação de Imagens

This work presents a \varphi-divergence, which is a generalization of Shannon and Tsallis relative entropies, as dissimilarity measure in image segmentation by clustering. The tests are performed by segmenting the text region of scanned images with noise from the NoisyOffice database. Based on the results, the proposed clustering method was more applicable than established methods, such as Otsu threshold and classical K-means.