节点文献
图像的压缩、加密和检索及其算法研究
A Study on Compression, Encryption and Retrieval of Image and Their Arithmetics
【作者】 黄艳;
【导师】 王以治;
【作者基本信息】 华中科技大学 , 计算数学, 2005, 硕士
【摘要】 目前,在通信、航天、教学等众多领域都迫切需要图像的有效压缩和高效传输; 为了防止在图像传输和共享过程中的侵权行为,迫切需要行之有效的加密技术; 为了提高在浩瀚的图像库中寻找特定图像的效率,迫切需要高效的图像检索方法。本文对图像的压缩、加密和检索及其算法实现进行了研究。这里在对图像进行小波包分解的基础上实现图像压缩。根据小波系数的分布特点严格推导得出:可以用能量和方差作为衡量小波系数重要性的标准,并据此对系数矩阵块进行选择分组,自适应地进行量化,而且不同的量化方式可以共用码表。本文利用数字水印实现图像的加密,并设计了两种水印算法,一种是结合JPEG 标准的量化特点设计的水印算法,它充分利用了图像的特征和人的视觉特性,水印的检测只需要原始图像一小部分的信息,在含水印图像遭受不同程度的压缩、剪切、模糊加噪等攻击后仍能有效地检测到水印的存在。另外一种是在充分研究图像的数据隐藏能力的基础上,提出利用边缘点数、分布均匀度和亮度来综合衡量图像的数据隐藏能力; 借用通迅中“调频”思想,提出了区别于传统水印技术的“调频水印”技术; 选择数据隐藏能力强的图像块的中频系数组成基带信号,结合调频思想进行水印的嵌入和提取。这种水印技术可兼顾不可见性和鲁棒性,并且可以实现高效的盲检测。最后本文提出了一种高级检索方式:基于姿态理解的人的检索,结合人的体形特点,对图中人的姿态进行理解,并用矢量来描述人的姿态,借助矢量映射来实现符合人的理解的高级检索。
【Abstract】 Currently, valid compression and efficient transfers of images are needed in communication, navigation and teaching etc; people require valid encryption method in order to prohibit infringing in the process of delivering and sharing of images; an efficiently retrieval method is urgently demanded when looking for a certain image in extensive image’s database. In this paper, image is compressed efficiently based on the wavelet packets decomposition. It is concluded that energy and variance can be serve as the criterions which can be used to judge the importance of the coefficients through strictly deduction based on the character of the wavelet coefficients, these coefficients can be classed according to the criterion, after that they should be quantified adaptively; further more, different methods of quantization can use the same code table. In this paper, two methods are used to encrypt images. One method is designing watermark algorithm according to the principle of the DCT transformation and JPEG standard, taking good use of the sense of visual feature of person and the characteristic of image. The examinations of watermark only need a very small part of data from the original image according to the principle of our algorithm. It is easy to examine the watermark’s existence when the image which contain watermark suffering from the compression of different degree, shearing to slice, and so on. Another method is based on the study of the ability of data conceals adequately, and proposed that the ability of data conceals can be scaled by using the degree of edge, the even degree of distribute and luminance. Using the idea of frequency modulation that belongs to communication engineering, and bringing up “frequency modulation watermark”which different to traditional watermark technique. Constituting base band signal by select the intermediate frequency coefficients of image piece which has strong data conceals ability, then, embedding and picking up watermark according to the idea of frequency modulation. This kind of watermark technique can give attention to invisibleness and robustness, and realize blind examination efficiently. A high-level retrieval method prompted in this paper is the retrieval of human based on the comprehension of pose. Comprehend the pose of human, descript the human by vector and realize the retrieval that tally with comprehension of human through the method of vector mapping warehouse.
【Key words】 image compression; digital watermarking; hiding capacity; pose understanding; retrieval vector;
- 【网络出版投稿人】 华中科技大学 【网络出版年期】2006年 05期
- 【分类号】TP309
- 【下载频次】279