• • 上一篇    下一篇

一种分块压缩感知观测值的图像篡改认证算法

李珊珊%周爽%蒋天发   

  • 基金资助:
    国家自然科学基金[40571128]、国家民委重点科研项目(MZY20004)

Image Tampering Authentication Algorithm based on Block Compressive Sensing Measurements

LI Shan-shan%ZHOU Shuang%JIANG Tian-fa   

  • About author:中南民族大学计算机科学学院,湖北武汉,430074%中南财经政法大学统计与数统学院,湖北武汉,430073

摘要: 文章针对压缩感知理论特点,提出一种新的算法。分块认证可将篡改定位到分块级别,具有较高的篡改定位准确度。算法首先将图像分成若干块,分块大小可以由篡改定位调整,然后对各个分块进行观测得到压缩的观测值。在观测域中将得到的观测值进行LDPC编码生成水印,并将水印嵌入到分块图像中。水印提取时同样对篡改后的图像进行随机投影,对提取的水印进行解码和比较,实现了图像的篡改检测与图像认证。根据实验研究证实,该算法对局部篡改的检测功能较强,算法实现简单且提取过程更加安全。

Abstract: Based on the theory of compressed sensing, a new algorithm is proposed. The tampering can be positioned to block-level based on block certification, and with high positioning accuracy. Algorithm firstly divides the image into several blocks, and block size can be changed from tampering with positioning, then observed value of each block to obtain the observations compression. In the observational domain, we will get the observed value of LDPC code to generation watermark, and the watermark is embedded into the block in the image. Watermark extraction, compressed sensing theory will be used random projection image, and decoding the extracted watermark. Compared with before, it has realized the image tampering detection and image authentication. According to the experimental result, the algorithm has the local tamper detection function, good concealment algorithm. This algorithm is simple and more secure.