Representing Symbolic Pictures Using Iconic Indexing
Chin-Chen Chang (National Chung Chen University, Taiwan), Yung-Kuan Chan (National Chung Chen University, Taiwan), Annie Y.H. Chou (National Chiao Tung University, Taiwan) and Wei-Pang Yang (National Chiao Tung University, Taiwan)
Copyright: © 2001
2D-string and 2D H-string are commonly used to describe the spatial relationships among the objects of a picture. This chapter introduces a new data structure, called N-string, to characterize the spatial information of the objects on a picture. The N-string is isomorphic to the 2D H-string in postfix notation. Algorithms for transformations between symbolic pictures and N-strings are developed as well. In addition, this chapter also provides an efficient method for image matching based on the N-string representation. This chapter shows that in most cases the storage space needed by N-strings is less than that by 2D-strings, too. Besides, the experimental results show that N-strings are superior to 2D-strings. Hence both N-string and 2D H-string in postfix notation are good for representing symbolic pictures. That is, both N-string and 2D H-string in postfix notation are good for iconic indexing of pictures.