Cooperative robust estimation using layers of support

摘要: We present an approach to the problem of representing images that contain multiple objects or surfaces. Rather than using edge-based represent segmentation a scene, we propose multilayer estimation framework which uses support maps image into homogeneous chunks. This support-based can are split disjoint regions, have surfaces transparently interleaved. Our is based on extension robust methods provide theoretical basis for estimation. use selection criteria derived from minimum description length principle decide how many in describing image. method has been applied number different domains, including decomposition range constituent objects, sequences higher-order motion fields, and separation tracked features distinct rigid-body motions. >