Using Hidden Markov Model In Hierarchical Image Mining

Study on image mining methods shows variety viewpoint of these methods to semantic recognition and perception of concepts in large image data bases.Decisive pattern is a combination of different low level feature values, which are unique and significant for describing a semantic concept. If an image includes patternof concept 1, surely it has concept1. Also if this image includes other patterns, so it has the concepts of thosepatterns. Decisive patterns of all the concepts of the hierarchical structure levels of data base are extracted; therefore all the concepts have unique patterns and rules. Rule bases obtainfrom combination of all of the rule base of each concept, then semantic concepts will classify using hierarchicalHidden Markov Model(HMM). Peruse on researches in image mining domain, demonstrates the challenge in the exact definition of image mining. One ofthe aims of this paper is to present a definition for image mining. Presented method can apply on multi concepts images, as inreality each image has some concepts. Experimental results on large databases with many concepts, shows that our approach is more effective than some previous ones

Keywords:

Image mining , Data mining , Knowledge discovery , Semantic gap , Semantic image classifier , Hierarchical classification , Semantic rule base