检索优化的媒体资源管理系统的设计与实现

发布时间：2019-07-08 15:13

【摘要】：伴随互联网技术的飞速发展,媒体产业积累的数字资产越来越多,媒体资源管理(DAM)成为媒体行业的一个研究热点。作为媒体资源管理的一个重要功能,基于内容的多媒体内容检索逐渐应用到DAM系统中。如何综合利用提取到的媒体素材的低层特征,获得更好的检索效果?针对这个问题,本文提出了一种融合全局特征和局部特征的特征融合模型,利用不同特征各自的优势,在此基础上设计并实现了检索效果优化的媒体资源管理系统。首先,本文以图像为例,研究了基于内容的图像检索中与图像特征提取的相关问题,提出了一种融合图像局部特征和全局特征的检索方法,其中局部特征以基于SIFT描述子的词袋模型对图像特征进行描述,全局特征以图像的颜色空间为例对图像进行描述,将该方法在Corel1000图像检索测试数据集上进行试验。实验表明,图像特征融合对于系统的整体检索准确率有一定提高。然后,本文对现有的DAM系统进行了对比研究,在充分考虑系统的扩展性、开放性以及灵活性等设计原则的基础上,通过分析系统功能模块和资产数据结构设计了 DAM系统,最后利用J2EE、MongoDB等技术实现了系统,完成了对系统检索功能的测试以及对测试结果的分析。通过对检索优化DAM系统的检索测试结果证实其适用性和可靠性。
文内图片：

图片说明：图２－ｌ邋ＨＳＶ颜色空间［２４】逡逑
[Abstract]:With the rapid development of Internet technology, more and more digital assets are accumulated in the media industry. Media resource management (DAM) has become a research hot spot in the media industry. As an important function of media resource management, content-based multimedia content retrieval is gradually applied to DAM system. How to make comprehensive use of the low-level features of the extracted media materials to obtain better retrieval results? In order to solve this problem, a feature fusion model combining global features and local features is proposed in this paper. Based on the advantages of different features, a media resource management system with optimal retrieval effect is designed and implemented. Firstly, taking the image as an example, this paper studies the problems related to image feature extraction in content-based image retrieval, and proposes a retrieval method for fusion image local features and global features, in which the local features describe the image features with the word bag model based on SIFT descriptor, and the global features take the color space of the image as an example to describe the image, and the method is tested on the Corel1000 image retrieval test data set. The experimental results show that the image feature fusion can improve the overall retrieval accuracy of the system to a certain extent. Then, this paper makes a comparative study of the existing DAM system. On the basis of fully considering the design principles of expansibility, openness and flexibility of the system, the DAM system is designed by analyzing the functional modules and asset data structure of the system. Finally, the system is realized by using J2EE, MongoDB and other technologies, and the test of the retrieval function of the system and the analysis of the test results are completed. The applicability and reliability of the retrieval optimization DAM system are verified by the retrieval test results.
【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP391.41;TP311.52

【参考文献】