Crimped mineral wools are characterized by a strongly anisotropic microstructure, whose local preferential orientation is highly heterogeneous. It is proposed to identify the local anisotropic elastic behavior through a combination of different tools based on image analysis. First, the local orientation map is determined from a reference image. Second, a series of images captured at different loading stages is analyzed with a digital image correlation code to estimate the displacement field. Last, an inverse analysis is applied to evaluate the four elastic moduli of the local elastic properties. This procedure is tested against a set of experimental data on mineral wool with different crimping structures.