Design And Implementation Of Indexing Mechanism For Image Information Based On The Berkeley DB

A large number of image file data comes about in the Internet environment。In order to provide the image file information retrieval and analysis, first of all we should solve the image file information retrieval's index storage problem, the existing picture search engine's searching are inefficiency and inaccuracy can't satisfied the user's need. This paper based on the text-based image information retrieval, researched the index mechanism of the retrieval process.The indexing mechanisms of text-based image information retrieval process can be divided into three parts: Index keyword extraction, indexing database storage, and search results sorting. This paper primarily focused on introducing the content of these three areas, and emphasized on the keyword extraction and indexing data store these two processes.This paper analyzed several classical dictionary mechanisms in the keyword extraction process, for the current dictionary mechanism structure is complicated and inefficient, it proposed an improved dictionary mechanism. Analysis based on the open source Berkeley DB, in order to further improve the indexing of data storage and retrieval efficiency and ameliorate the Hash access method of Berkeley DB, this paper proposed a no-split bucket allocation mechanism, reducing the expensive overhead resulting from the split process. Integrated the database indexing techniques and image information retrieval techniques, this paper designed and implemented an efficient indexing mechanism, and the experimental results show that the new index mechanism has promoted image information retrieval efficiency for the large amount of data stored.