基于Beamlet的抗干扰的设计图检索

发布时间：2018-05-07 07:43

本文选题：基于内容的图像检索 + beamlet分析　；参考：《河北师范大学》2013年硕士论文

【摘要】：近几十年来，随着互联网的普及和数字图像处理技术的成熟，基于内容的图像检索成为信息时代的研究热点．互联网上设计图的数量大量增加，如服装设计图、提包设计图、首饰设计图，还有卡通设计图等等，为了合法的保护设计师的知识产权，方便高效的进行设计图检索成为基于内容的图像检索的重要应用前景之一．实际得到的图像常常会受到噪声污染，或人为或图像传输设备造成，因此抗干扰的设计图检索的研究是必要的．本文的主要工作是：首先，对多尺度几何分析—beamlet分析进行了深入研究，详细介绍了beamlet基本理论．目前为止从查阅的文献看，只有对beamlet算法结果的描述，而没有对算法的具体实现步骤作系统的阐述，，本文的重点是给出了beamlet算法—无结构算法和树结构算法的具体实现步骤．其次，用beamlet算法实现了强噪声下弱信号的提取，在高斯噪声、瑞利噪声和周期噪声中都能很好的恢复几乎完整的原直线的存在方式；同时还用beamlet算法实现了图像线特征的多尺度描述和图像区域的提取．最后，也即本文重点，在基于内容的图像检索中，尤其是设计图检索，提出了beamlet特征，无论图像含不含噪声，beamlet算法可直接提取图像的线特征，不需要对图像降噪预处理．本文初步实现了基于beamlet的图像检索，检索结果表明基于beamlet的抗干扰的设计图检索的有效性，为beamlet变换的应用扩展了一个新方向．
[Abstract]:In recent decades, with the popularity of the Internet and the maturity of digital image processing technology, content-based image retrieval has become a research hotspot in the information age. The number of designs on the Internet has increased substantially, such as clothing drawings, bag drawings, jewelry drawings, cartoon designs, etc., in order to legally protect the intellectual property rights of designers, Convenient and efficient design map retrieval is one of the important applications of content-based image retrieval. The actual images are often polluted by noise or caused by artificial or image transmission equipment, so it is necessary to study the anti-jamming design map retrieval. The main work of this paper is as follows: firstly, the multi-scale geometric analysis -Beamlet analysis is deeply studied, and the basic theory of beamlet is introduced in detail. So far, only the description of the results of the beamlet algorithm, but not the detailed implementation steps of the algorithm is described. The emphasis of this paper is to give the implementation steps of the beamlet algorithm, the unstructured algorithm and the tree structure algorithm. Secondly, beamlet algorithm is used to extract the weak signal under strong noise. In Gao Si noise, Rayleigh noise and periodic noise, the almost complete original line can be recovered. At the same time, the beamlet algorithm is used to realize the multi-scale description of image line features and the extraction of image region. Finally, that is the focus of this paper, in the content-based image retrieval, especially the design image retrieval, the beamlet feature is proposed. Regardless of whether the image contains noiseless features, the line feature of the image can be extracted directly without the need of image de-noising preprocessing. In this paper, the image retrieval based on beamlet is preliminarily implemented. The retrieval results show that the anti-jamming design map retrieval based on beamlet is effective, which extends a new direction for the application of beamlet transform.
【学位授予单位】：河北师范大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP391.41

【参考文献】