Research And Application Of Content-based Image Retrieval

With the development of multimedia and internet technology,the resources of image and video are increaseing plentfulness.Nowadays it is a cloud computing and big data era,and the change of the information between intenet is more frequently.The image data with its characteristics of intuitive and attractive are full of internet.Such as the data in museum,the medical images,the records of crim,the remote sensing image,the monitor in traffic,the shopping in internet,the digital library and UAV.As the disappear of paper media,the eletronic media rise quickly,the effective management of image data and faster image reseache become an issue needing to be researched and solved.As a reseach target,this paper design a content-based image retrieval.At the first,the paper analyisised the colour features,texture features,shape features and spatial information of image which is based on vision.And then,the research embraced the characteristics of image,the way of feature extraction and the scope of application.At the same time,the thesis researched the way and prospect of application in the content-based image retrieval.Through building the model of database,the system completed the feature extraction about the colour space,markov texture features and edge feature.The eigenvalue to be geted were put into database.For every different kind of image feature,the system gave a kind of matching algorithm.For colour space,system used the Euclidean distance to compute the similar distance between the different colours,the^??17?distance for markov texture features and edge feature.In the end,thesis designed the holistic content-based image retrieval,putting into use the feature extraction and the matching algorithm.Experiment proved that the system has batter effect of retrieval and definite basis of apply.