Design And Implentation Of Media Asset Management System With Optimized Retrieval Approach

With the rapid development of internet technology,the digital asset collected in media industry has been increasing to a huge amount,thus,Digital Asset Management(DAM)becomes a hotspot in media and science area.As a significant function in DAM,Content-based Media Retrieval has been applied to DAM system.How should we leverage invluable low-level features of media source to gain retreival result with high quality?To solve the issue,we proposed a feature fusion model,which combines holistic features and local ones synthesizing each advantages respectively to design and implement a DAM system with an optimized retrieval engine.Initially,we took image assets as an example to study issues related to Content-based Image Retrieval,specifically the feature extraction.We proposed a retreival method merging local and holistic image features,in which the local features were displayed as Bag-of-Word model based on SIFT discriptors,and the holistic features were represented by HSV color space vectors.After that,we did the experiment on the image dataset Corel1000 that proved the feature fusion method could wholly enhance the accuracy of retrieval to some extent.Then,we investigated and compared current DAM systems,and then designed the DAM system after adquate consideration on the system's expansibility,openness,and flexibility.Finally,we utilized technique such as J2EE,MongoDB,etc.to implement the system,test and analyse its function.We validated the system's applicability and reliability according to the retrieval test results.