摘要: This contribution presents a new approach to the recognition of priori known 3-D objects in single 2-D images. The underlying model is embedded domain CAD-based vision using viewer-centered generate set normalized views. They serve as basis for an optimal selection properties features. aspect idea used grouping values into aspect-trees. aim this identify correct view object seen image and thereby distinguish between different objects. achieved with appropriate traversion aspect-trees which proves ability interpretation system Aspik recognize complex environments robustly efficiently. time complexity recognizing nonoccluded O (n2 � m), where n number considered m