A Fuzzy Set Representation Approach for Object-based

摘要

The objective of this work is to automatically retrieve from a large number of images for a specified object autocorrelation features. In order to cope with the ambiguity of spatial relative position concepts, we propose a fuzzy set method for object-based image retrieval to represent the spatial relationships of different objects. The proposed representation approach improves the traditional "all-or-nothing" definition method, in two aspects. First, the approach strengthens the discrimination power of different objects by comparing an object to a fuzzy set representing the degree of position satisfaction to a reference object. Second, to achieve better intra-class invariance power, Fuzzy Attributed Relational Graphs (FARGs) are used in this framework, where each node represents an image object and each edge represents a relation between two objects. The testing on two different data sets shows that the fuzzy set approach can partially illustrate visual differences of images of the same class and deliver satisfactory retrieval results of relevant images of different classes.