Study On Content-based Retrieval Of Video Of Digital Asset Management In Printing Company

Digital Asset Management (DAM) is a new technology. It is appeared in abroad in1990s. Digital asset acquisition, storage and re-use are the targets of DAM. DAM is mainlyused in the news media, book publishing, entertainment services, government medicaland other rich media industry. Digital asset is a form of digital assets by definition. Inprinting industry, DAM is a very important source. Every day, every company iscontinuously generating various types of electronic documents (including all kinks of videofiles, graphics, images, multimedia files and other files). These files are various kinds andlarge quantities, but they are prototypes of digital assets of companies. There are twokinds of assets which are tangible assets and intangible assets. Digital assets arevaluable intangible assets.Now, the internet is full of a variety of publishing, printing and printingmachinery-related videos. We call them printing videos. It is no doubt that these printingvideos provide a great convenience for the people whose work in a printing industry. But,at the same time, these video files bring some troubles to the workers, too. According tothe data of Seybol in2000, there are40%of the files is a duplicate in company storage,and30%of them can be reused by a small change. The data also point out that thesearch for files number is75of a creative staff in a week, per people per week spent3hours on data management, and files transfer takes3hours, too. Thus, centralized digitalasset management can improve productivity and reduce costs.Face to the massive video data on the network, how can we find the video clips whichwe need quickly and accurately? This requires the support of effective video retrievaltechnology. The traditional text-based video retrieval method can not meet the needs ofpeople, and content-based video retrieval has become a new trend in video retrieval.This paper is mainly about the key technologies of content-based video retrieval(CBVR). Shot detection and key-frame extraction are included.In the field of printing, the scenes of printing video are relatively single, and videocontents are relatively simple. The content of printing video is almost introducing themachine running and operation and other relatively simple scenes. This kind of videoalways has some phenomenon, such as uneven illumination and camera with shake.Therefore, this paper chooses H component of HSV color model.Because of gradual shot has various types and complicated process, in this paper, a new method is proposed. This method is very effective.This paper proposes a novel approach of key-frame extraction based on averageframe-difference. The experimental results indicate that the new approach is very effectiveand efficient for extracting key-frames of printing video.