基于超像素分割与三维空间的图像哈希  

Image Hash Based on Super-Pixel Segmentation and Three-Dimensional Space

作　　者：刘帅 赵琰 LIU Shuai;ZHAO Yan(College of Electronics and Information Engineering,Shanghai University of Electric Power,Shanghai 200120,China)

机构地区：[1]上海电力大学电子与信息工程学院,上海200120

出　　处：《计算机仿真》2022年第12期264-270,共7页Computer Simulation

基　　金：国家自然科学基金资助项目(61802250);上海市科委部分地方院校能力建设项目(20020500700)。

摘　　要：为了提高图像哈希算法的分类性能并减小哈希序列的长度,采用超像素分割和三维空间构建图像哈希序列。首先对图像进行预处理,提取图像亮度分量Y并对Y分量进行分块,再对每块进行超像素分割,对获得的最大超像素块的位置信息用凸包算法提取其二维空间轮廓特征;同时对亮度分量Y构建三维空间并提取三维空间距离特征;将二维空间轮廓特征和三维空间距离特征结合起来构成中间哈希序列。最后利用密钥对中间哈希序列置乱后得到图像哈希序列。仿真结果表明,算法具有良好的鲁棒性、区别性和安全性,与现有的一些算法相比,具有更短的哈希序列,同时有更好的图像分类性能。In order to improve the classification performance of the image hashing algorithm and reduce the hash length, the image hashing sequence is constructed by using superpixel segmentation and three-dimensional space. Firstly, the image is pre-processed. The luminance component Y is extracted and divided into non-overlapped blocks. Each small block is super-pixel segmented and extracted contour feature. Then the position information of the largest part of each small block is extracted. At the same time, the luminance component Y is constructed in three-dimensional space and the three-dimensional spatial distance feature is extracted. The intermediate hash sequence is formed by combining the two-dimensional space contour feature and the three-dimensional space distance feature. Finally, the intermediate hash sequence is scrambled to obtain the final hash sequence. Experimental results show that the proposed algorithm has good robustness, discriminability and key security. Compared with some existing algorithms, the algorithm has shorter hash sequences and better image classification performance.

关 键 词：图像哈希 位置信息 超像素分割 凸包算法 三维空间 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]