The Representation, Retrieval, and Matching of Symbolic Images. Based on Object Attributes and Spatial Relationships

      Аннотация: A new framework is proposed in this book. This framework supports the representation, retrieval, and matching of images in a symbolic image database.For representation of symbolic images, models SRG and SRGT are proposed. For similarity retrieval of symbolic images with higher effectiveness, a method CPM incorporating a new data structure CP_DAG is proposed. For the matching of two SRG or labeled graphs, the proposed algorithm CISC (Class-wise ISomorphism testing with limited baCktracking) utilizes the feature of labeled graphs to compactly encode the vertex invariants into vertex and graph signatures and performs signature comparisons in linear time.

Ключевые слова: Algorithm, symbolic image, similarity retrieval, spatial relationship, isomorphism testing