云概念相似性度量及其在数字水印中的应用研究
|
摘要
不确定性知识表示和处理,包括定性定量转换、软计算、变粒度计算等,逐渐成为网络计算中亟待解决的热点问题。双向认知计算模型——云模型是李德毅院士在结合概率论和模糊数学理论两者的基础之上,通过赋予样本点以随机确定度统一刻画概念的随机性、模糊性及其关联性。不确定性知识的表示和处理在人工智能、数据挖掘、机器学习等领域能有广泛应用。在运用云模型的时候,对定量的数据,往往使用云变换将其转换为定性概念,通过定性概念的相似性度量,来表征数据的相似性。在云数字水印技术方面,云概念间的相似性度量分析的作用尤为明显,例如对提取嵌入到宿主对象的水印云滴后,将会判断提取的云水印是否就是嵌入到宿主的水印云滴。云概念相似性度量算法的好坏影响到云模型在运用中的效率和精度,因此研究云概念间的相似性有一定的理论价值和实际意义,也是对云模型研究的扩展。
本文的主要工作包含以下几个方面：
1)介绍了不确定计算的相关知识以及双向认知计算模型——云模型。
2)正态云概念的相似性度量已有多种算法被提出,分析了现有云概念间的相似性度量算法,存在如下不足：时间复杂度高、对数字特征值要求过高、不能较好推广至高阶、区分度低等。在现有研究成果基础上,推广了期望曲线的定义,推导了云概念间每阶期望曲线间的距离公式,再通过加权求平均距离来得到两个正态云概念的相似度,提出了基于多阶期望曲线加权的云概念相似性度量算法。
3)将不确定性数学模型引入数字水印领域,把关注点放在研究水印的产生、提取,以及验证云水印的存在。将二阶正态云的数字特征值作为密钥,通过正向云变换产生一维正态水印云滴。在变换域,将载体图像块进行DCT变换,通过对中频系数的调制,把水印云滴嵌入到中频系数中。通过不同逆向云变换算法对带有云水印的载体图像进行水印提取,分析其误差大小。鲁棒性攻击实验中,运用本文的相似性度量算法与原始嵌入的云水印进行比较,实验结果表明,将具有不确定性的云模型相关理论引入到数字水印领域是可行的,其优势在于部分水印云滴能较好还原到原始云概念的三个参数。
Uncertain knowledge representation and processing is becoming a hot issue to be solved in the network computing, including qualitative and quantitative conversion, soft computing, variable granular computing and so on. Bidirectional cognition calculated model (cloud model) is that Academician Li D.Y described the randomness, fuzziness and relevance of concept on randomly determined degree by the given sample points, which based on both of the probability theory and the fuzzy mathematical theory. Uncertain knowledge representation and processing is widely used in the field of artificial intelligence, data mining, machine learning. When using the cloud model, for the quantitative data, we often convert it to the qualitative concept by cloud method, then characterize the similarity of data with the similarity measure of qualitative concept. Analyzing the similarity measure of cloud concept plays an important role, especially in the cloud digital watermarking. For example, after extracting the watermark cloud droplets embedded into the host, it will judge the extracted cloud watermark that whether is the watermark droplets embedded into the host or not. The quality of cloud concept similarity measure algorithm affects the efficiency and accuracy in the use of cloud theory,. There is some theoretical value and practical significance in studying the similarity of cloud concept,. It's also the extended research of the cloud theory.
The main content in this paper includes the following aspects:
1) Describe the research background and significance, as well present the related knowledge of the uncertainty calculate and bidirectional cognition calculated model (cloud model).
2) A variety of normal cloud concept similarity measure algorithms have been proposed. Analyzing the existing cloud concept similarity measure algorithms, there are the following deficiencies:high time complexity, exigent requirements of digital eigen values, worse extension to high order, low discrimination and so on. According to the existing theory, this chapter promotes the definition of the expectation curve, and derives range formula between each order expectation curve of cloud concept. Then it gains the similarity values of two normal cloud concepts by weighting the average distance. This paper desires a cloud concept similarity measure algorithm based on multi-order expectation curve weighted.
3) The uncertainty mathematical model is introduced into the field of digital watermarking. The concerns focus on the generation and extraction of the watermark, and how to verify the presence of the cloud watermark. The digital eigen values of the second-order normal cloud as the cipher code, by means of the forward cloud transformation that generated one-dimensional normal cloud watermark. Vector image block is transformed by DCT transformation in the transform domain, then the watermark cloud droplets are embedded into the middle frequency coefficients through modulating the middle frequency coefficients. It analyzes the error of the watermarks that are extracted from the vector image with cloud watermark by the different reverse cloud transformation algorithm. In the attacking robustness experiment, the similarity measure algorithm in this paper is compared with the original one-dimensional normal cloud watermark algorithm. The experiment results indicate that introducing the correlation theory of the uncertainty cloud model theory to the field of digital watermarking is feasible. Its advantage is that part of the watermark cloud droplets can be better resorted to the three parameters of the original cloud concept.
本文的主要工作包含以下几个方面：
1)介绍了不确定计算的相关知识以及双向认知计算模型——云模型。
2)正态云概念的相似性度量已有多种算法被提出,分析了现有云概念间的相似性度量算法,存在如下不足：时间复杂度高、对数字特征值要求过高、不能较好推广至高阶、区分度低等。在现有研究成果基础上,推广了期望曲线的定义,推导了云概念间每阶期望曲线间的距离公式,再通过加权求平均距离来得到两个正态云概念的相似度,提出了基于多阶期望曲线加权的云概念相似性度量算法。
3)将不确定性数学模型引入数字水印领域,把关注点放在研究水印的产生、提取,以及验证云水印的存在。将二阶正态云的数字特征值作为密钥,通过正向云变换产生一维正态水印云滴。在变换域,将载体图像块进行DCT变换,通过对中频系数的调制,把水印云滴嵌入到中频系数中。通过不同逆向云变换算法对带有云水印的载体图像进行水印提取,分析其误差大小。鲁棒性攻击实验中,运用本文的相似性度量算法与原始嵌入的云水印进行比较,实验结果表明,将具有不确定性的云模型相关理论引入到数字水印领域是可行的,其优势在于部分水印云滴能较好还原到原始云概念的三个参数。
Uncertain knowledge representation and processing is becoming a hot issue to be solved in the network computing, including qualitative and quantitative conversion, soft computing, variable granular computing and so on. Bidirectional cognition calculated model (cloud model) is that Academician Li D.Y described the randomness, fuzziness and relevance of concept on randomly determined degree by the given sample points, which based on both of the probability theory and the fuzzy mathematical theory. Uncertain knowledge representation and processing is widely used in the field of artificial intelligence, data mining, machine learning. When using the cloud model, for the quantitative data, we often convert it to the qualitative concept by cloud method, then characterize the similarity of data with the similarity measure of qualitative concept. Analyzing the similarity measure of cloud concept plays an important role, especially in the cloud digital watermarking. For example, after extracting the watermark cloud droplets embedded into the host, it will judge the extracted cloud watermark that whether is the watermark droplets embedded into the host or not. The quality of cloud concept similarity measure algorithm affects the efficiency and accuracy in the use of cloud theory,. There is some theoretical value and practical significance in studying the similarity of cloud concept,. It's also the extended research of the cloud theory.
The main content in this paper includes the following aspects:
1) Describe the research background and significance, as well present the related knowledge of the uncertainty calculate and bidirectional cognition calculated model (cloud model).
2) A variety of normal cloud concept similarity measure algorithms have been proposed. Analyzing the existing cloud concept similarity measure algorithms, there are the following deficiencies:high time complexity, exigent requirements of digital eigen values, worse extension to high order, low discrimination and so on. According to the existing theory, this chapter promotes the definition of the expectation curve, and derives range formula between each order expectation curve of cloud concept. Then it gains the similarity values of two normal cloud concepts by weighting the average distance. This paper desires a cloud concept similarity measure algorithm based on multi-order expectation curve weighted.
3) The uncertainty mathematical model is introduced into the field of digital watermarking. The concerns focus on the generation and extraction of the watermark, and how to verify the presence of the cloud watermark. The digital eigen values of the second-order normal cloud as the cipher code, by means of the forward cloud transformation that generated one-dimensional normal cloud watermark. Vector image block is transformed by DCT transformation in the transform domain, then the watermark cloud droplets are embedded into the middle frequency coefficients through modulating the middle frequency coefficients. It analyzes the error of the watermarks that are extracted from the vector image with cloud watermark by the different reverse cloud transformation algorithm. In the attacking robustness experiment, the similarity measure algorithm in this paper is compared with the original one-dimensional normal cloud watermark algorithm. The experiment results indicate that introducing the correlation theory of the uncertainty cloud model theory to the field of digital watermarking is feasible. Its advantage is that part of the watermark cloud droplets can be better resorted to the three parameters of the original cloud concept.
引文
[1]于少伟,曹凯,赵模.基于一维云模型的交通信息预测算法[J].山东大学学报,2007,37(2)：123-126.
[2]邵帅,梁少卫.云模型在水资源供求预测中的应用[J].西北民族大学学报(自然科学版),2008,29(3)：80-82.
[3]唐桂军,白翠翠,阎显.云模型在商品月销售量预测中的应用[J].科学技术与工程,2010,10(1)：241-243.
[4]周辅疆,朱小冬,程永伦.云模型在训练弹药消耗预测中的应用研究[J].舰船电子工程,2009,29(10)：58-59.
[5]黄海生,王汝传.基于隶属云理论的主观信任评估模型研究[J].通信学报,2008,29(4)：16-19.
[6]范定国,贺硕,段富,牛保宁.一种基于云模型的综合评判模型[J].科学情报开发与经济,2003,13(12)：157-159.
[7]柴日发,徐文骞,曾文华.基于云模型的BP算法改进[J].计算机仿真,2002,19(3)：123-125.
[8]张光卫,康建初,李鹤松等.基于云模型的全局最优化算法[J].北京航空航天大学学报,2007,33(4)：487-490.
[9]阎芳,刘丙午.基于云模型的动态物流过程知识表示[J].物流技术,2008,27(6)：75-76.
[10]王洪利,冯玉强.基于云理论的群体复杂决策中不确定知识的表示[J].黑龙江大学自然科学学报,2007,24(3)：342-344.
[11]李众,杨一栋.一种新的基于二维云模型不确定性推理的智能控制器[J].控制与决策,2005,20(8)：867-872.
[12]张飞舟,范跃祖,沈程智,李德毅,陈晖.利用云模型实现智能控制倒立摆[J].控制理论与应用,2000,17(4)：519-523.
[13]马鸿耀.粒计算与云模型在彩色图像分割中的应用研究[D].重庆：重庆邮电大学学报,2012.
[14]马鸿耀,王国胤,张清华.基于云模型的多粒度彩色图像分割[J].计算机工程,2012,20(38)：184-187.
[15]许凯,秦昆,黄伯和,杜鹢.基于云模型的图像区域分割方法[J].中国图像图形学报,2010,15(5)：757-763.
[16]秦昆,李振宇,李辉等.基于云模型和格网划分的图像分割方法[J].测绘通报,2006,5：1-7.
[17]李万臣,葛秘蕾,井志强.基于云理论的图像分割新方法[J].应用科技,2010,37(3)：45-48.
[18]许凯,秦昆,裴韬.一种交互式的云模型图像分割方法[J].计算机工程与应用,2006,34：33-36.
[19]赵卫伟,李德毅.基于云模型的入侵检测方法[J].计算机工程与应用,2003,39(26)：158-160,164.
[20]孟详怡,张光卫,刘常昱等.基于云模型的主观信任管理模型研究[J].系统仿真学报,2007,19(14)：3310-3317.
[21]路峰,吴慧中.网络环境下基于云模型的信任评估与决策方法研究[J].系统仿真学报,2009,21(2)：421-426.
[22]黄海生,王汝传.基于隶属云理论的主观信任评价评估模型研究[J].通信学报,2008,29(4)：13-19.
[23]高键,姜长生,李众.一种新的云模型控制器设计[J].信息与制,2005,34(2)：159-161.
[24]赵永辉,史浩山.基于云模型的无线传感器网络拥塞及速率控制策略[J].传感技术学报,2010,23(1)：134-138.
[25]维特根斯坦著,陈嘉映译.哲学研究(Philosophical Investigation) [M].上海：上海人民出版社,2001.
[26]薛忆沩.概念隐喻理论与汉文化研究[J].深圳大学学报.2002(2).
[27]Lakoff G, Johnson M. Metaphors We Live By. Chicago:Univ[J].1980.
[28]Lakoff G. The contemporary theory of metaphor[J]. Metaphor and thought,1993,2: 202-251.
[29]MITCHELL, T.M. and T.M. MITCHELL. Machine Learning[M].1997.
[30]张勇,赵东宁,李德毅.数字水印技术及进展[J].解放军理工大学学报,2003,4(3)：1-5.
[31]Simmons G J. The prisoners'problem and the subliminal channel[C]. Advances in Cryptology. Springer US,1984:51-67.
[32]Bloom J A, Cox I J, Kalker T, et al. Copy protection for DVD video[J]. Proceedings of the IEEE,1999,87(7):1267-1276.
[33]Cox I J, Linnartz J P M G. Some general methods for tampering with watermarks [J]. Selected Areas in Communications, IEEE Journal on,1998,16(4):587-593.
[34]Zhao J, Koch E, Luo C. In business today and tomorrow[J]. Communications of the ACM,1998,41(7):67-72.
[35]Cox I J, Miller M L, Bloom J A. Watermarking applications and their properties[C]. Information Technology:Coding and Computing,2000. Proceedings. International Conference on. IEEE,2000:6-10.
[36]蒋伟德.基于Web-Service的电子票据数字水印主动控制模型[J].计算机应用软件,2006,(10).
[37]李开乐.相似离度及其使用技术[J].气象学报,1986,44(2)：174-182.
[38]罗阳,赵伟.相似性度量研究及最有相似系数[J],China Academic Journal Electronic Publishing House All rights reserved,1994-2011.
[39]张勇,赵东宁,李德毅.相似云及其度量分析方法[J].信息与控制,2004,33(2)：130-132.
[40]张光卫,李德毅,李鹏等.基于云模型协同过滤推荐算法[J].软件学报,2007,18(10)：2403-2411.
[41]李海林,郭崇慧,邱望仁.正态云模型相似度计算方法[J].电子学报,2011,11(39)：2561-2567.
[42]Neisser U. Cognitive psychology[M]. New York:Appleton-Century-Crofts,1967.
[43]ZAJONC R B, MORRISSETTE J. The role of uncertainty in cognitive change[J]. Journal of abnormal and social psychology,1960,61:168.
[44]Quayle D, Ball L. Cognitive uncertainty in syllogistic reasoning:An alternative mental models theory[C]. The twenty-third annual conference of the cognitive science society, Mahwah, New Jersey, Lawrence erlbaun associates,2001:67-71.
[45]王梓坤.概率论基础及其应用[M].北京,北京师范大学出版社,1996：99-112.
[46]Zadeh L A. Information and control[J]. Fuzzy sets,1965,8(3):338-353.
[47]Pawlak Z. Rough sets[J]. International journal of computer & information sciences, 1982,11(5):341-356.
[48]王国胤,姚一豫,于洪.粗糙集理论与应用研究综述[J].计算机学报,2009,32(7)：1229-1246.
[49]Wang G Y, Rough set based uncertainty knowledge expressing and processing [C], RSFDGrC 2011, Moscow, Russia,2011:11-18.
[50]李德毅,杜鹢.不确定性人工智能[M].北京,国防工业出版社,2005：135-138.
[51]王国胤,李德毅,姚一豫,梁吉业等.云模型与粒计算[M].北京,科学出版社,2012：15-27.
[52]Wang G Y, Xu C L. Cloud model-A bidirectional cognition model between concepts extension and intension with Lixto[C]. In first international conference AMLTA 2012, Cairo, Egypt,2012:391-400.
[53]许昌林,王国胤.实现稳定双向认知映射的逆向云变换算法[J].模式识别.已录用 待发表.
[54]李德毅,刘常昱.论正态云模型的普适性[J].中国工程科学,2004,6(8)：28-34.
[55]刘禹,李德毅,张光卫等.云模型雾化特性及在进化算法中的应用[J].电子学报,2009,37(8)：1651-1658.
[56]王国胤,许昌林,张清华,王晓蓉.p阶正态云模型的递归定义及其统计分析[J].计算机学报,待发表.
[57]许昌林,张清华,王晓蓉.p阶正态云模型的逆向云算法[J].计算机学报,待发表.
[58]王立新.正态云的基本数学性质与云滤波器[J].个人通信,2011,24(5)：19-25.
[59]刘玉超.基于云模型的粒计算方法研究[D].北京,清华大学,2012：45-49.
[60]延臣,王劲峰.基于云模型的图像分割方法研究[M],北京,地质出版社,2003：67-73.
[61]MovieLens[EB/OL].http://movielens.unn.edu,1997-9-19/1988-5-22.
[62]张飞.基于云模型的文本数字水印算法研究与实现[J].电子设计工程,2010,18(5)：8-11.
[63]王飞燕,李峰,陈桂松.基于一维正态云模型的半脆弱文本水印[J].计算机工程与设计,2008,29(17)：4578-4580.
[64]徐敏,王衍波,李涛.基于云模型的XML数字水印算法.军事通信技术[J].2009,30(1)：41-45.
[65]张菁,杨静宇.一维正态云模型水印研究[J].武汉理工大学学报.2010,32(16)：48-51.
[2]邵帅,梁少卫.云模型在水资源供求预测中的应用[J].西北民族大学学报(自然科学版),2008,29(3)：80-82.
[3]唐桂军,白翠翠,阎显.云模型在商品月销售量预测中的应用[J].科学技术与工程,2010,10(1)：241-243.
[4]周辅疆,朱小冬,程永伦.云模型在训练弹药消耗预测中的应用研究[J].舰船电子工程,2009,29(10)：58-59.
[5]黄海生,王汝传.基于隶属云理论的主观信任评估模型研究[J].通信学报,2008,29(4)：16-19.
[6]范定国,贺硕,段富,牛保宁.一种基于云模型的综合评判模型[J].科学情报开发与经济,2003,13(12)：157-159.
[7]柴日发,徐文骞,曾文华.基于云模型的BP算法改进[J].计算机仿真,2002,19(3)：123-125.
[8]张光卫,康建初,李鹤松等.基于云模型的全局最优化算法[J].北京航空航天大学学报,2007,33(4)：487-490.
[9]阎芳,刘丙午.基于云模型的动态物流过程知识表示[J].物流技术,2008,27(6)：75-76.
[10]王洪利,冯玉强.基于云理论的群体复杂决策中不确定知识的表示[J].黑龙江大学自然科学学报,2007,24(3)：342-344.
[11]李众,杨一栋.一种新的基于二维云模型不确定性推理的智能控制器[J].控制与决策,2005,20(8)：867-872.
[12]张飞舟,范跃祖,沈程智,李德毅,陈晖.利用云模型实现智能控制倒立摆[J].控制理论与应用,2000,17(4)：519-523.
[13]马鸿耀.粒计算与云模型在彩色图像分割中的应用研究[D].重庆：重庆邮电大学学报,2012.
[14]马鸿耀,王国胤,张清华.基于云模型的多粒度彩色图像分割[J].计算机工程,2012,20(38)：184-187.
[15]许凯,秦昆,黄伯和,杜鹢.基于云模型的图像区域分割方法[J].中国图像图形学报,2010,15(5)：757-763.
[16]秦昆,李振宇,李辉等.基于云模型和格网划分的图像分割方法[J].测绘通报,2006,5：1-7.
[17]李万臣,葛秘蕾,井志强.基于云理论的图像分割新方法[J].应用科技,2010,37(3)：45-48.
[18]许凯,秦昆,裴韬.一种交互式的云模型图像分割方法[J].计算机工程与应用,2006,34：33-36.
[19]赵卫伟,李德毅.基于云模型的入侵检测方法[J].计算机工程与应用,2003,39(26)：158-160,164.
[20]孟详怡,张光卫,刘常昱等.基于云模型的主观信任管理模型研究[J].系统仿真学报,2007,19(14)：3310-3317.
[21]路峰,吴慧中.网络环境下基于云模型的信任评估与决策方法研究[J].系统仿真学报,2009,21(2)：421-426.
[22]黄海生,王汝传.基于隶属云理论的主观信任评价评估模型研究[J].通信学报,2008,29(4)：13-19.
[23]高键,姜长生,李众.一种新的云模型控制器设计[J].信息与制,2005,34(2)：159-161.
[24]赵永辉,史浩山.基于云模型的无线传感器网络拥塞及速率控制策略[J].传感技术学报,2010,23(1)：134-138.
[25]维特根斯坦著,陈嘉映译.哲学研究(Philosophical Investigation) [M].上海：上海人民出版社,2001.
[26]薛忆沩.概念隐喻理论与汉文化研究[J].深圳大学学报.2002(2).
[27]Lakoff G, Johnson M. Metaphors We Live By. Chicago:Univ[J].1980.
[28]Lakoff G. The contemporary theory of metaphor[J]. Metaphor and thought,1993,2: 202-251.
[29]MITCHELL, T.M. and T.M. MITCHELL. Machine Learning[M].1997.
[30]张勇,赵东宁,李德毅.数字水印技术及进展[J].解放军理工大学学报,2003,4(3)：1-5.
[31]Simmons G J. The prisoners'problem and the subliminal channel[C]. Advances in Cryptology. Springer US,1984:51-67.
[32]Bloom J A, Cox I J, Kalker T, et al. Copy protection for DVD video[J]. Proceedings of the IEEE,1999,87(7):1267-1276.
[33]Cox I J, Linnartz J P M G. Some general methods for tampering with watermarks [J]. Selected Areas in Communications, IEEE Journal on,1998,16(4):587-593.
[34]Zhao J, Koch E, Luo C. In business today and tomorrow[J]. Communications of the ACM,1998,41(7):67-72.
[35]Cox I J, Miller M L, Bloom J A. Watermarking applications and their properties[C]. Information Technology:Coding and Computing,2000. Proceedings. International Conference on. IEEE,2000:6-10.
[36]蒋伟德.基于Web-Service的电子票据数字水印主动控制模型[J].计算机应用软件,2006,(10).
[37]李开乐.相似离度及其使用技术[J].气象学报,1986,44(2)：174-182.
[38]罗阳,赵伟.相似性度量研究及最有相似系数[J],China Academic Journal Electronic Publishing House All rights reserved,1994-2011.
[39]张勇,赵东宁,李德毅.相似云及其度量分析方法[J].信息与控制,2004,33(2)：130-132.
[40]张光卫,李德毅,李鹏等.基于云模型协同过滤推荐算法[J].软件学报,2007,18(10)：2403-2411.
[41]李海林,郭崇慧,邱望仁.正态云模型相似度计算方法[J].电子学报,2011,11(39)：2561-2567.
[42]Neisser U. Cognitive psychology[M]. New York:Appleton-Century-Crofts,1967.
[43]ZAJONC R B, MORRISSETTE J. The role of uncertainty in cognitive change[J]. Journal of abnormal and social psychology,1960,61:168.
[44]Quayle D, Ball L. Cognitive uncertainty in syllogistic reasoning:An alternative mental models theory[C]. The twenty-third annual conference of the cognitive science society, Mahwah, New Jersey, Lawrence erlbaun associates,2001:67-71.
[45]王梓坤.概率论基础及其应用[M].北京,北京师范大学出版社,1996：99-112.
[46]Zadeh L A. Information and control[J]. Fuzzy sets,1965,8(3):338-353.
[47]Pawlak Z. Rough sets[J]. International journal of computer & information sciences, 1982,11(5):341-356.
[48]王国胤,姚一豫,于洪.粗糙集理论与应用研究综述[J].计算机学报,2009,32(7)：1229-1246.
[49]Wang G Y, Rough set based uncertainty knowledge expressing and processing [C], RSFDGrC 2011, Moscow, Russia,2011:11-18.
[50]李德毅,杜鹢.不确定性人工智能[M].北京,国防工业出版社,2005：135-138.
[51]王国胤,李德毅,姚一豫,梁吉业等.云模型与粒计算[M].北京,科学出版社,2012：15-27.
[52]Wang G Y, Xu C L. Cloud model-A bidirectional cognition model between concepts extension and intension with Lixto[C]. In first international conference AMLTA 2012, Cairo, Egypt,2012:391-400.
[53]许昌林,王国胤.实现稳定双向认知映射的逆向云变换算法[J].模式识别.已录用 待发表.
[54]李德毅,刘常昱.论正态云模型的普适性[J].中国工程科学,2004,6(8)：28-34.
[55]刘禹,李德毅,张光卫等.云模型雾化特性及在进化算法中的应用[J].电子学报,2009,37(8)：1651-1658.
[56]王国胤,许昌林,张清华,王晓蓉.p阶正态云模型的递归定义及其统计分析[J].计算机学报,待发表.
[57]许昌林,张清华,王晓蓉.p阶正态云模型的逆向云算法[J].计算机学报,待发表.
[58]王立新.正态云的基本数学性质与云滤波器[J].个人通信,2011,24(5)：19-25.
[59]刘玉超.基于云模型的粒计算方法研究[D].北京,清华大学,2012：45-49.
[60]延臣,王劲峰.基于云模型的图像分割方法研究[M],北京,地质出版社,2003：67-73.
[61]MovieLens[EB/OL].http://movielens.unn.edu,1997-9-19/1988-5-22.
[62]张飞.基于云模型的文本数字水印算法研究与实现[J].电子设计工程,2010,18(5)：8-11.
[63]王飞燕,李峰,陈桂松.基于一维正态云模型的半脆弱文本水印[J].计算机工程与设计,2008,29(17)：4578-4580.
[64]徐敏,王衍波,李涛.基于云模型的XML数字水印算法.军事通信技术[J].2009,30(1)：41-45.
[65]张菁,杨静宇.一维正态云模型水印研究[J].武汉理工大学学报.2010,32(16)：48-51.