AFFINE INVARIANT TEXTURE SEGMENTATION AND SHAPE FROM TEXTURE BY VARIATIONAL-METHODS

We address the problem of texture segmentation by rising a novel affin e invariant model. The introduction of affine invariance as a requirem ent for texture analysis goes beyond what is known of the human perfor mance and also beyond the psychophysical theories. We propose to compu te texture features using affine invariant intrinsic neighborhoods and affine invariant intrinsic orientation matrices. We discuss several p ossibilities for the definition of the channels and give comparative e xperimental results where an affine invariant Mumford-Shah type energy functional is used to compute the multichannel affine invariant segme ntation. We prove that the method is able to retrieve faithfully the t exture regions and to recover the shape from texture information in im ages where several textures are present. The numerical algorithm is mu ltiscale.