Research And Application Of Facial And Textual Information Retrieval Based On FRBR

For the information retrieval, it includes not only retrieval techniques, but also datacollection，data pretreatment，data structure，data storage，description of informationretrieval and relevance calculation of object query. The article discusses the techniqueproblems in depth about how to retrieve the face image information and textinformation.The article gives a comprehensive analysis of FRBR model, then mainlyresearches the key retrieval techniques about data structure and data retrieval fusion.The main research work and contributions of this dissertation are as follows.1) The article analyses FRBR model in depth, and sets the feasibility of organizingdata with it.2) The article analyses rules of features information describing and achieves howto organize features effectively with FRBR; a distributed cooperative informationretrieval framework for data fusion is putted forward, and the rationality and validity arevalidated experimentally based on mass data.3) According to requirements, the article puts forward a method of face detectionwith gray projection algorithm and template matching method, and features areorganized with FRBR.Experimental results on a set of images in ORL show that thismethod can locate the human face region rapidly and effectively.4) A powerful word library is set up, and a method for word-segment is proposed,features are organized with FRBR. It adapts to the changes of various documents andcan meet the requirements of text feature extraction.5) The theory of FRBR is applied into a floating population service andmanagement information system, and validity of retrieval for fusing image and textinformation is validated based on it.Experimental results show that information retrievalfor fusing image and text information can be realized according to the theory of FRBR,and the accuracy of results will be improved.