基于H.265的部分视频加密技术
发布时间：2019-02-11 09:07
【摘要】：视觉作为人类认知世界的最直观的方式。数字视频紧随着IT的浪潮,获得非常迅速的发展。但是,一个完整的视频是由个大量的数据信息组成,而视频要想获得实际的应用,就必须对视频进行压缩编码,现如今发布的较为成熟的编码标准是由ITU-T和ISO/IEC联合发布的H.265/HEVC,它与前代H.264/AVC相比,能够用比其少50%的码率达到一样的视频质量,它还拥有并行处理的设计,其软硬件实现将具有更为出色的实用性,因此它的出现将迅速取代H.264/AVC,并在视频业务中受到广泛追捧,但是它的出现并没有改变视频信息频繁遭到恶意的攻击和破坏的命运,因此基于H.265/HEVC的视频加密技术也越来越受到关注。近年来,研究学者们设计出了许多的加密算法,最早采用全加密的方法,虽然它的安全性能较高,但是,由于它巨大的计算工作量,使得它的实用性就变弱了。后来又有基于DCT系数加密、熵编码过程加密等与编解码过程相结合的加密算法,这类算法虽然在一定程度上减少了加密的数据量,但对于保密的安全系数也相对降低了,随着这类的研究的不断深入,为了减少计算量的同时能拥有可观的加密等级,并且不影响压缩效率,本文提出了一种全新的加密算法,其工作内容和目的如下:对H.265的预测、变换、量化等部分的了解和分析,学习现有的一些加密算法,比较其优劣和计算复杂度,在此基础上,我们希望设计出一种能实现视频的部分加密的算法,同时在对密钥未知的情况下使得图像在解码端能够被重建,但是重建后的图像质量有所下降,使其不能获得原始的信息,从而达到加密目的。相比于现在已经有的在熵编码阶段或者比特流中加密的加密方法,我们在文章中提出了在变换阶段进行加密:根据改变蝶形中第二阶段的旋转角度,从而设计出一系列变换矩阵,并通过仿真验证这些矩阵是否会影响编码效率,然后通过RC4算法随机产生的密钥对一系列新的变换矩阵进行选择,从而替代原始的DCT,因此在解码端对密钥未知的情况下,对于采用的具体哪一个变换矩阵就是未知的,从而实现视频的部分加密的目的。文章最后将我们的算法应用到具体的视频序列中去,然后针对密钥信息掌握的不同程度来分析解码端的视频质量,对加密等级进行评估。
[Abstract]:Vision is the most intuitive way for human beings to perceive the world. Digital video is following the tide of IT, obtain very rapid development. However, a complete video is composed of a large amount of data information, and if the video is to be used in practice, it must be compressed and encoded. The more mature coding standard released today is H.265 / HEVC, released jointly by ITU-T and ISO/IEC, which can achieve the same video quality with 50 percent less bit-rate than the previous H.264/AVC. It also has the design of parallel processing, and its hardware and software implementation will be more practical, so it will quickly replace H.264% AVC, and be popular in the video business. However, its appearance has not changed the fate that video information is frequently attacked and destroyed by malicious attacks. Therefore, video encryption technology based on H.265/HEVC has been paid more and more attention. In recent years, researchers have designed many encryption algorithms, the earliest full encryption method, although its security performance is high, but because of its huge computing workload, it becomes less practical. Then there are encryption algorithms based on DCT coefficient encryption, entropy encoding process encryption and encoding and decoding process. Although these algorithms reduce the amount of encrypted data to a certain extent, the security factor for security is also relatively reduced. With the development of this kind of research, a new encryption algorithm is proposed in this paper in order to reduce the amount of computation and have a considerable encryption level without affecting the compression efficiency. Its contents and purposes are as follows: understanding and analyzing the prediction, transformation, quantization and other parts of H.265, learning some existing encryption algorithms, comparing their advantages and disadvantages and computational complexity, on this basis, We hope to design a partial encryption algorithm to realize the video, and at the same time, we can reconstruct the image at the decoding end without knowing the key, but the quality of the reconstructed image is degraded, which makes it unable to obtain the original information. In order to achieve the purpose of encryption. Compared with the existing encryption methods used in entropy encoding or bitstream encryption, we propose to encrypt in the transformation phase by changing the rotation angle of the second stage in the butterfly. A series of transformation matrices are designed and verified by simulation, and then a series of new transformation matrices are selected by random key generated by RC4 algorithm to replace the original DCT,. Therefore, when the key is unknown at the decoding end, which transformation matrix is unknown, so that the purpose of partial encryption of video can be realized. Finally, our algorithm is applied to the specific video sequence, and then the video quality of the decoder is analyzed according to the different degree of key information, and the encryption level is evaluated.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN919.81
本文编号：2419584
[Abstract]:Vision is the most intuitive way for human beings to perceive the world. Digital video is following the tide of IT, obtain very rapid development. However, a complete video is composed of a large amount of data information, and if the video is to be used in practice, it must be compressed and encoded. The more mature coding standard released today is H.265 / HEVC, released jointly by ITU-T and ISO/IEC, which can achieve the same video quality with 50 percent less bit-rate than the previous H.264/AVC. It also has the design of parallel processing, and its hardware and software implementation will be more practical, so it will quickly replace H.264% AVC, and be popular in the video business. However, its appearance has not changed the fate that video information is frequently attacked and destroyed by malicious attacks. Therefore, video encryption technology based on H.265/HEVC has been paid more and more attention. In recent years, researchers have designed many encryption algorithms, the earliest full encryption method, although its security performance is high, but because of its huge computing workload, it becomes less practical. Then there are encryption algorithms based on DCT coefficient encryption, entropy encoding process encryption and encoding and decoding process. Although these algorithms reduce the amount of encrypted data to a certain extent, the security factor for security is also relatively reduced. With the development of this kind of research, a new encryption algorithm is proposed in this paper in order to reduce the amount of computation and have a considerable encryption level without affecting the compression efficiency. Its contents and purposes are as follows: understanding and analyzing the prediction, transformation, quantization and other parts of H.265, learning some existing encryption algorithms, comparing their advantages and disadvantages and computational complexity, on this basis, We hope to design a partial encryption algorithm to realize the video, and at the same time, we can reconstruct the image at the decoding end without knowing the key, but the quality of the reconstructed image is degraded, which makes it unable to obtain the original information. In order to achieve the purpose of encryption. Compared with the existing encryption methods used in entropy encoding or bitstream encryption, we propose to encrypt in the transformation phase by changing the rotation angle of the second stage in the butterfly. A series of transformation matrices are designed and verified by simulation, and then a series of new transformation matrices are selected by random key generated by RC4 algorithm to replace the original DCT,. Therefore, when the key is unknown at the decoding end, which transformation matrix is unknown, so that the purpose of partial encryption of video can be realized. Finally, our algorithm is applied to the specific video sequence, and then the video quality of the decoder is analyzed according to the different degree of key information, and the encryption level is evaluated.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN919.81
【参考文献】
相关期刊论文 前1条
1 ;Parallel processing architecture of H.264 adaptive deblocking filters[J];Journal of Zhejiang University(Science A:An International Applied Physics & Engineering Journal);2009年08期
，本文编号：2419584
本文链接：https://www.wllwen.com/kejilunwen/xinxigongchenglunwen/2419584.html
