基于数据降维与对称二值模式的图像Hash算法

发布时间：2018-01-28 01:00

本文关键词： 图像处理图像Hash 数据投影降维对称局部二值模式量化规则一维组合混沌映射决策阈值　出处：《激光与光电子学进展》2017年02期 　论文类型：期刊论文

【摘要】：为了解决当前图像Hash算法难以兼顾较高的感知稳健性与篡改识别率的不足,提出了基于数据投影降维机制与对称局部二值模式的紧凑图像Hash算法。利用双线性插值来预处理图像,使Hash具有固定的长度;引入对数极坐标变换,将其转变为二次图像;利用Gabor滤波器平滑二次图像;基于模糊集理论,设计对称局部二值模式算子,获取稳健特征;定义数据投影降维机制与量化规则,生成紧凑的中间Hash比特序列;构造一维组合混沌映射,建立加密模型,完成比特序列扩散,以生成图像Hash;并引入汉明距离,估算初始图像与接收端图像的Hash相似度,联合决策阈值,完成图像认证。测试数据表明,与当前图像Hash技术相比,该算法的Hash更紧凑,且其感知稳健性与敏感性更高。
[Abstract]:In order to solve the current image Hash algorithm is difficult to take into account the high perceptual robustness and tamper recognition rate deficiencies. A compact image Hash algorithm based on the dimensionality reduction mechanism of data projection and symmetric local binary mode is proposed. The bilinear interpolation is used to preprocess the image so that the Hash has a fixed length. The logarithmic polar coordinate transformation is introduced to transform it into a quadratic image. The quadratic image is smoothed by Gabor filter. Based on fuzzy set theory, symmetric local binary pattern operators are designed to obtain robust features. Data projection dimensionality reduction mechanism and quantization rules are defined to generate compact intermediate Hash bit sequences. One dimensional combined chaotic map is constructed, encryption model is established, bit sequence diffusion is completed, and the image Hashis is generated. The hamming distance is introduced to estimate the Hash similarity between the initial image and the receiver image, and the decision threshold is combined to complete the image authentication. The test data show that compared with the current image Hash technology. The Hash of the algorithm is more compact, and its perceptual robustness and sensitivity are higher.
【作者单位】：平顶山教育学院计算机系;郑州大学信息工程学院;
【基金】：河南省科技计划重点项目(102102210416) 河南省软科学研究计划项目(152400410323)
【分类号】：TP391.41
【正文快照】： 随着计算机技术的日益完善,各种图像编辑软件也不断升级,其编辑功能越来越强大,可以对图像进行021004-1任意修改,而不留下任何篡改痕迹,仅凭人眼无法识别图像的真伪,严重威胁着图像信息安全[1-2]。为了应对这种图像篡改,有效识别图像信息的真伪,国内外学者提出了诸多图像内容

【相似文献】