The Study of Image Feature Extraction and Retrieval Based on Granular Computing
Zhou Yong-jie · Computer Knowledge and Technology · 2010
Apply the HSV model lining with visual sensors,anti-geometric theory of deformation of the wavelet transform and Hu Moments theory to Respectively extract the image features of color,texture and shape.Then combine granular computing theory and the three characteristics to granulate the three characteristics.This way,we can get three feature vectors and normalize them.Based on that,we can set the weight of each image feature,then get the images' close degree by means of similarity measurement algorithm and associate relevance feedback mechanism with that to adjust each of the weights.Repeat the adjusting process until the users are satisfied,then the image retrieval that integrate the color,texture and shape is realized.