An automatic method to determinate the degree of flocculence of a galaxy

We propose a new method to determine the flocculence of a galaxy image. Flocculence is characterized by a texture feature computed using a bank of Gabor filters. These filters, inspired by the human visual system, uniformly cover the spatial-frequency domain. Texture features are obtained by extracting statistics from sub-windows in the filtered images. Flocculent regions are then detected using a machine learning approach. First results are presented on the EFIGI dataset.