The project falls into the area of image analysis and object recognition. We mainly focus on two important connected areas: object description by invariant features and image fusion. The theoretical concept is based on the formulation of all tasks as optimization problems and looking for proper cost functions and algorithms for their maximization. We propose a representation of objects and images by projections onto a conveniently chosen functional bases. Unlike fixed bases, we propose to adapt the basis functions according to the particular image and task. This allows to create new invariant representations with high discriminability and specificity. In the image fusion part, we use the above representations to design probabilistic fusion methods which are able to handle heavily degraded images and to work under real-life uncertain conditions. In addition to 2D and 3D images, we plan to work also with vector/flow fields. The project assumes also practical applications, namely in classification of biological shapes, in industrial vision, and in face recognition. The The project assumes also practical applications, namely in classification of biological shapes, in industrial vision, and in face recognition. The project goal is to design adaptive invariant features for image representation and to use them for developing methods for high-level image fusion, which should be able to handle significantly distorted input images and and uncertain conditions.