基于边缘映射抽样与多维尺度压缩的紧凑图像哈希算法
|
摘要
为了充分利用图像的显著特征来生成鲁棒哈希,并提高其生成效率,提出基于边缘映射抽样与多维尺度压缩的紧凑图像哈希算法。联合双线性插值与高斯低通滤波,对初始图像进行预处理操作;引入双阈值Canny算子,提取预处理图像的显著边缘映射。将预处理图像与边缘映射分割为一系列的非重叠子块;计算每个子块的像素值总和,找出其中元素值最大的子块,从中进行选择性抽样,确定对应的包含结构信息最丰富的子块。基于SVD(Singular Value Decompostion)机制,将子块的奇异值视为鲁棒特征;利用Fourier变换与残差机制,提取预处理图像的全局显著特征。将两种特征组合,形成中间哈希序列,再引入多维尺度机制压缩获取紧凑哈希,并通过Logistic映射来实施加密,形成最终的哈希。测试数据表明:较已有的哈希方案而言,该方案具备更高的鲁棒性与哈希生成效率,呈现出更好的ROC曲线。
In order to make full use of the salient features of the image to generate robust hashes and improve the efficiency of its generation, we proposed a compact image hashing algorithm based on edge map sampling and multidimensional scale compression in this paper. Combining bilinear interpolation and Gaussian low-pass filtering, the initial image was preprocessed. We introduced the dual threshold canny operator to extract the salient edge mapping of preprocessed images. The preprocessed image and the edge mapping were divided into a series of non-overlapping sub-blocks. We calculated the sum of pixel values of each edge mapping sub-block to form a transition matrix, and the sub-block with the largest element value was found. Selective sampling was carried out from the preprocessed image to determine the corresponding sub-block with the richest structural information. Based on SVD mechanism, these sampling sub-blocks were processed and their singular values were regarded as robust features. We adopted the Fourier transform and residual mechanism to extract local salient features of preprocessed images. The intermediate hash sequence was formed by combining the two features. And the multidimensional scale was introduced to complete the compression and obtain the compact hash. The compact hash was encrypted by Logistic map to form the final hash. Test results show that this scheme has higher robustness and hash generation efficiency, as well as better ROC curves than the existing hash scheme.
In order to make full use of the salient features of the image to generate robust hashes and improve the efficiency of its generation, we proposed a compact image hashing algorithm based on edge map sampling and multidimensional scale compression in this paper. Combining bilinear interpolation and Gaussian low-pass filtering, the initial image was preprocessed. We introduced the dual threshold canny operator to extract the salient edge mapping of preprocessed images. The preprocessed image and the edge mapping were divided into a series of non-overlapping sub-blocks. We calculated the sum of pixel values of each edge mapping sub-block to form a transition matrix, and the sub-block with the largest element value was found. Selective sampling was carried out from the preprocessed image to determine the corresponding sub-block with the richest structural information. Based on SVD mechanism, these sampling sub-blocks were processed and their singular values were regarded as robust features. We adopted the Fourier transform and residual mechanism to extract local salient features of preprocessed images. The intermediate hash sequence was formed by combining the two features. And the multidimensional scale was introduced to complete the compression and obtain the compact hash. The compact hash was encrypted by Logistic map to form the final hash. Test results show that this scheme has higher robustness and hash generation efficiency, as well as better ROC curves than the existing hash scheme.
引文
[1] 金晓民, 张丽萍. 混合特征与颜色矢量角度的图像哈希认证算法[J]. 计算机科学与探索, 2018, 38(7): 1102-1115.
[2] 张勇, 黄家荣. 非负矩阵分解耦合环形分割的图像哈希认证算法[J]. 计算机工程与设计, 2017, 38(9): 2464-2471.
[3] 王彦超, 郭静博, 周丽宴. 基于数据投影降维机制与对称局部二值模式的紧凑图像哈希算法[J]. 激光与光电子学进展, 2017, 54(2): 4-16.
[4] Sun R, Zeng W J. Secure and robust image hashing via compressive sensing [J]. Multimedia Tools and Applications, 2014, 70(3): 1651-1665.
[5] 王彦超. 基于联合特征与中心方向信息的图像哈希算法[J]. 西南大学学报(自然科学板), 2018, 40(2): 113-124.
[6] 唐振军, 杨帆, 黄紫晴. 基于PCA特征距离的图像哈希算法[J]. 广西师范大学学报(自然科学版), 2016, 34 (4): 9-18.
[7] Tang Z J, Chen L, Zhang X Q. Robust image hashing with tensor decomposition [J]. IEEE Transactions on Knowledge & Data Engineering, 2018, 121(99): 1-12.
[8] Choi Y S, Park J H . Image hash generation method using hierarchical histogram[J]. Multimedia Tools and Applications, 2012, 61(1):181-194.
[9] 段军, 高翔. 基于统计滤波的自适应双阈值改进canny算子边缘检测算法[J]. 激光杂志, 2015, 36(1): 10-13.
[10] Karsh R K, Laskar R H. Robust image hashing through DWT-SVD and spectral residual method[J]. EURASIP Journal on Image and Video Processing, 2017, 22(1): 31-45.
[11] Li J Z, Yu C Y. Color image watermarking scheme based on quaternion Hadamard transform and Schur decomposition[J]. Multimedia Tools & Applications, 2017, 77(5): 1-17.
[12] 赵爱罡, 王宏力, 杨小冈. 基于非线性局部滤波的红外小目标检测方法[J]. 工程科学学报, 2016, 38(11): 1652-1658.
[13] Kumar S, Pant M, Kumar M. Colour image segmentation with histogram and homogeneity histogram difference using evolutionary algorithms[J]. International Journal of Machine Learning & Cybernetics, 2018, 9(1): 163-183.
[14] Davarzani R, Mozaffari S, Yaghmaie K . Perceptual image hashing using center-symmetric local binary patterns[J].Multimedia Tools and Applications,2016,75(8):4639-4667.
[15] Machado J A T, Din? Erdal, Baleanu D . Analysis of UV spectral bands using multidimensional scaling[J]. Signal Image & Video Processing, 2015, 9(3):573-580.
[16] Yu J, Li Y, Xie X. Image encryption algorithm by using the logistic map and discrete fractional angular transform[J]. Optica Applicata, 2017, 47(1): 141-155.
[17] Schaefer G, Stich M. UCID: An uncompressed color image database[C]// Proceedings of SPIE, Storage and Retrieval Methods and Applications for Multi-media,2004:472-480.
[18] Petitcolas F A P. Watermarking schemes evaluation [J]. IEEE Signal Proc Magazine, 2000, 17(5): 1-4.
[2] 张勇, 黄家荣. 非负矩阵分解耦合环形分割的图像哈希认证算法[J]. 计算机工程与设计, 2017, 38(9): 2464-2471.
[3] 王彦超, 郭静博, 周丽宴. 基于数据投影降维机制与对称局部二值模式的紧凑图像哈希算法[J]. 激光与光电子学进展, 2017, 54(2): 4-16.
[4] Sun R, Zeng W J. Secure and robust image hashing via compressive sensing [J]. Multimedia Tools and Applications, 2014, 70(3): 1651-1665.
[5] 王彦超. 基于联合特征与中心方向信息的图像哈希算法[J]. 西南大学学报(自然科学板), 2018, 40(2): 113-124.
[6] 唐振军, 杨帆, 黄紫晴. 基于PCA特征距离的图像哈希算法[J]. 广西师范大学学报(自然科学版), 2016, 34 (4): 9-18.
[7] Tang Z J, Chen L, Zhang X Q. Robust image hashing with tensor decomposition [J]. IEEE Transactions on Knowledge & Data Engineering, 2018, 121(99): 1-12.
[8] Choi Y S, Park J H . Image hash generation method using hierarchical histogram[J]. Multimedia Tools and Applications, 2012, 61(1):181-194.
[9] 段军, 高翔. 基于统计滤波的自适应双阈值改进canny算子边缘检测算法[J]. 激光杂志, 2015, 36(1): 10-13.
[10] Karsh R K, Laskar R H. Robust image hashing through DWT-SVD and spectral residual method[J]. EURASIP Journal on Image and Video Processing, 2017, 22(1): 31-45.
[11] Li J Z, Yu C Y. Color image watermarking scheme based on quaternion Hadamard transform and Schur decomposition[J]. Multimedia Tools & Applications, 2017, 77(5): 1-17.
[12] 赵爱罡, 王宏力, 杨小冈. 基于非线性局部滤波的红外小目标检测方法[J]. 工程科学学报, 2016, 38(11): 1652-1658.
[13] Kumar S, Pant M, Kumar M. Colour image segmentation with histogram and homogeneity histogram difference using evolutionary algorithms[J]. International Journal of Machine Learning & Cybernetics, 2018, 9(1): 163-183.
[14] Davarzani R, Mozaffari S, Yaghmaie K . Perceptual image hashing using center-symmetric local binary patterns[J].Multimedia Tools and Applications,2016,75(8):4639-4667.
[15] Machado J A T, Din? Erdal, Baleanu D . Analysis of UV spectral bands using multidimensional scaling[J]. Signal Image & Video Processing, 2015, 9(3):573-580.
[16] Yu J, Li Y, Xie X. Image encryption algorithm by using the logistic map and discrete fractional angular transform[J]. Optica Applicata, 2017, 47(1): 141-155.
[17] Schaefer G, Stich M. UCID: An uncompressed color image database[C]// Proceedings of SPIE, Storage and Retrieval Methods and Applications for Multi-media,2004:472-480.
[18] Petitcolas F A P. Watermarking schemes evaluation [J]. IEEE Signal Proc Magazine, 2000, 17(5): 1-4.