Study On The Document Management System With The Capacity Of Searching For Massive Drawings

With the development of computer hardware the processing power of computersbecomes more and more powerful People can use computers to achieve greater number oftasks which are more complex Nowadays the information technology being highlydeveloped various applications emerge one after another and are developing towards thedirection of diversification and integration As one of the application softwares theenterprise document management system also has a lot of new changes with thedevelopment of computers Take the distributed document management system wedeveloped for example there are many new features in addition to the basic functions ofenterprise document management so as to meet the new requirements of modernizedelectronic document management such as search of drawing documents which requires thesearching function based on drawing contents And this is the focus of this paperThis paper is based on the actual project needs from problem-solving reality focusingon the research of content-based image retrieval (abbreviated as CBIR) which is currentlya hot spot at home and abroad In particular along with the development of digitaltechnology multimedia technology computer and network technology people are exposedto more and more image information Thus how to effectively organize manage andretrieve large image database becomes an urgent problem to address which is the researchcontent of the content-based image retrieval The early image retrieval systems such asKodak Picture Exchange System (KPX) are based on keyword retrieval method, that isthrough descriptive text retrieval to achieve the text-marked image retrieval In general thissearch technology often can not meet people's needs Since the image information is rich incontent in many cases it is difficult to fully describe with just a few keywords In additionthe selection of image features keywords is also very subjective Thus the content-basedimage retrieval technology emerged The so-called content-based image retrieval is a kindof search of various features based directly on the content of image media objects which can find the images with specified features or content directly from the image libraryTherefore to conduct image retrieval it is necessary to deeply study on the principlesof content-based retrieval Through the study of content-based retrieval the establishmentof CBIR functional module is achieved The CBIR module includes the following aspects:user interface namely human-machine interface used for the display of input sampleimages as well as the output display of approximate matching images imagepreprocessing here to conduct the enhanced process, binarization denoising and otheroperations of images (including the candidate images) feature extraction mainly extractionof image (also including the candidate images) features such as image statistics matchingoperation that is similarity calculation of the sample images and candidate images usingappropriate measurement method filtration and further processing of the results primarilyfurther processing of the results that meet the requirements such as sequence of the results(by up / down) so that users can find the target images more easilyThis paper is carried out around the above aspect I studied various methods of imagematching At present there are two commonly used global template matching and featurematching After the study of image feature extraction (such as color features texturefeatures shape features spatial relations features) it is found that actually thecharacteristics of the drawings are complex and the images cannot be matched with certainkind of specific feature Meanwhile taking into account the demand for high matchingaccuracy required in the project I finally chose the global template matching As to thesimilarity calculation in image matching I chose the classic Euclidean distance andnormalized cross-correlation measurement formula But in the system test in order to betteranalyze the results of image matching I also slightly improved the Euclidean distanceformula In addition to improve the image processing accuracy I researched the imagepreprocessing aspects of the drawing features for this project on gray-scale imagetransformation image binarization processing the threshold determination in binarizationand a variety of de-noising processing filtering features and the use limitations have alsobeen explored to find the suitable approach for this image retrieval functionHowever if truly want to realize the image search function it is also necessary tounderstand digital image programming so I further studied some basic theories of digital image programming and researched the images lattice indication the realization principleof digital images as well as storage formats in the computer of digital images And on thisbasis I defined a device-independent bitmap DIB class in order to facilitate programimplementationFinally through the full study on relevant technical knowledge of content-based imageretrieval based on the corresponding theory I designed the suitable search structure ofdrawings for this project and further refined each piece of functionality and technicalimplementation details in the design finally completed the implementation of the imagesearch function and integrated this functionality into our document management systemsuccessfully achieved customer requirements after testing the system as a wholeThe realization of drawings search function is my exploratory of content-based imageretrieval The content-based image retrieval technique, since the date it was raised hasdeveloped greatly and all aspects of researches are endless However trapped in the earlydevelopment of computer hardware the practical applications of this technology is notparticularly wide Scientific progress makes our tools more powerful Many new researchresults can be quickly transformed into real life So I think strengthening research in thisarea the establishment of image retrieval system with high efficiency and accuracy hasimportant practical significance and functional value...