Research On Image Retrieval Method Based On Multi-Feature

With the rapid development of network technology,information sharing has become an indispensable part in people's work,study and life.As an intuitive and lifelike visual information carrier,digital images are more popular to be acquired and shared in the iterative process of software and hardware technology,and the proportion of their data in the network is growing with each passing day.In order to find the parts of interest from a large number of image data for further processing accurately and quickly,it is an urgent problem to improve the efficiency of image retrieval.Text-based Image Retrieval(TBIR),due to its artificial intervention and description differences and other shortcomings,has gradually dropped out of the historical stage in the field of image retrieval,so Content-based Image Retrieval(CBIR)has gradually become a hot topic in the field of image retrieval.CBIR uses color,texture,shape,position,orientation and other features to describe the image,and through the similarity measurement of these features to complete the purpose of image retrieval.In this paper,the author propose an image retrieval method based on the fusion of global features and corner features.In order to further improve the efficiency of image retrieval,Genetic Algorithm is introduced into the process of image retrieval.Finally,the author designed an image retrieval system based of the features,to display the results of image retrieval clearly.The main research is as follows:1.An image retrieval method combining global and corner features is proposed.Firstly,the HSV histogram feature and LBP feature are extracted from the whole image,then the Hu invariant moment feature and the texture feature based on Gray-Level Co-occurrence Matrix are extracted from the corner of the image.Finally,the two kinds of features are fused together,and the similarity is measured by the relative Manhattan distance,and on the basis of these works,the author will complete the image retrieval.Experimental results show that this image retrieval method has a better effect.2.The author introduce Genetic Algorithm into image retrieval process.All of the extracted image features are normalized,and then these adaptive feature selection and weight setting are carried out by Genetic Algorithm simulating Darwin's theory of biological evolution.Through selection,crossover and mutation of the coded initial population from generation to generation,the individual with the best performance is evolved,and the optimal feature subset and weight coefficient are obtained by decoding the individual.Finally,the image retrieval was completed.Experimental data show that the efficiency of image retrieval is improved by using Genetic Algorithm for feature selection and weight setting.3.The author designed an image retrieval system based of the features,so the results of the image retrieval algorithm can be clearly presented.