Font Size: a A A

Based On Web Image Data Mining Research And Application

Posted on:2008-05-23Degree:MasterType:Thesis
Country:ChinaCandidate:Z Q LvFull Text:PDF
GTID:2178360242973195Subject:Computer technology
Abstract/Summary:PDF Full Text Request
The World Wide Web is a distributed global information resource containing a large amount of data relevant to essentially all domains of human activity. Given the high rate of the volume of data available on the WWW, finding useful information in such a large amount of data becomes a more difficult process every day. Data Mining is the term given to the automated discovery of non-obvious, potentially useful and previously unknown information from large data sources. So obtaining valuable information by Data Mining techniques intelligently and automatically, improving efficiency of the WWW has tremendous application values.Data mining provides us with a new way to resolve the problem that we can not make best of increasing huge data. Today, the main data source of data mining is historical databases which mainly include text and numerical data. The data mining on image data on www, especially image, lacks enough research. In this paper, we discussed some basal problems about image data mining, including definition, the oriels and technologies that image data mining needs, system structure and model.This paper has also studied based on the content picture retrieval technology (CBIR) the characteristic and the way, the CBIR technology mainly carries on the retrieval based on the picture content characteristic and the subject object characteristic; Discussed aspect the and so on picture color, texture, shape, spatial relations characteristic; Proposed the present picture excavation main technology, including the classification, gathers kind, the connection rule and the nerve network; Summarized the image data excavation key point and the difficulty lies in the characteristic withdrawing with the picture expression way, the picture comparison strategy, the machine learning and the picture automatic sign note, the retrieval system evaluation criteria and so on several aspects.The union based on the Web data mining method and based on the content picture retrieval, this article has conducted some research and the exploration to on the WWW picture information excavation and the retrieval system, has analyzed the Web picture search engine principle, the picture division method and the MPEG-7 standards, for realized this article to design the system to build the foundation and to construct a frame.Through the above content analysis and the research, constructs based on the Web image data excavation system which this article designs, first analyzed the system to need the function, built the rationale for the system realization, proposed the multi-dimensional data cube in the image data excavation concrete realization method, in view of the image data and the complexity, simplified the multi-dimensional data cubic structure; Uses the arithmetic of ultimate presenting times, enhanced the system computation rule speed, has designed amore complete system model, has drawn up various functions module diagram, and has introduced the sketching the desired object on the screen in the system, strengthened the system serviceability.
Keywords/Search Tags:Data Mining, CBIR, Information Retrieval on Internet, image
PDF Full Text Request
Related items