一种用于DEM数据认证与篡改定位的感知哈希算法

A Perceptual Hash Algorithm for DEM Data Authentication and Tamper Localization

DEM数据作为重要的基础地理信息数据,其数据完整性问题不容忽视。基于DEM数据完整性认证的要求,以及相关认证算法的欠缺,本文运用感知哈希技术设计了一种DEM数据认证算法,并可实现篡改定位。因DEM数据具有数据量大、细节丰富的特点,首先对其进行规则格网划分,将其划分为互不重叠的格网单元;然后对格网单元数据进行DCT分解,提取数据的特征信息以生成特征向量矩阵,并对特征向量矩阵进行摘要化处理;随后,使用Logistic混沌系统对简化后的特征向量矩阵进行置乱;对置乱矩阵进行量化、编码后,便可生成感知哈希序列。在数据认证时,首先计算原始数据与待验证数据的高程相对中误差,再将二者的感知哈希序列进行归一化汉明距离度量,结合判定阈值,即可对DEM数据进行数据认证与篡改定位。该算法对DEM数据的格式转换、水印嵌入等攻击有较强的鲁棒性,对各类改变内容的操作具有敏感性,并可实现DEM数据微小篡改的识别与定位。与已有的DEM完整性认证方法相比,将DEM数据的"内容"作为完整性度量的重要标准,在具体应用中更具有实用价值。

As a type of fundamental and important geographic data, the integrity of DEM data cannot be ignored.The commonly used technology for data integrity authentication is mainly based on traditional cryptography and digital watermarking technology. The former is very sensitive to the change of every bit of data, suitable for accurate authentication of text data;while latter is mostly based on data carrier for authentication, seldom considers if DEM data content changes or not, and needs additional secure channels and communication media.In this paper, based on the requirement of authenticity and integrity of DEM data and the shortcomings of related authentication algorithms, a DEM data authentication algorithm was designed based on the Perceptual Hashing technology, which can achieve tamper localization. Perceptual hashing is a kind of method that maps multimedia data unidirectionally into perceptual summary sets(i.e. hash sequences). It inherits the characteristics of traditional Hash functions such as unidirectionality, anti-collision, and summarization, and is robust to the operation of content preservation, so it can better meet the requirements of DEM data authentication. The main ideas of this algorithm are as follows: Based on the characteristics of a large amount of DEM data and abundant details, the DEM data is divided into regular and non-overlapping grids. Feature extraction is the key of Perceptual Hashing algorithm. In this paper, the discrete cosine transform was used to extract features and generate the eigenvector matrix. Then the eigenvector matrix was digested. Next, the simplified eigenvector matrix was scrambled by using a Logistic chaotic system to meet the security requirements of authentication.Followingly, the scrambled matrix was quantized and coded to generate perceptual hash sequence. In the data authentication stage, the relative error of elevation between the original data and the data to be validated was calculated firstly. Subsequently, the perceptual hash sequence of the original data and the data to be validated was normalized to measure the Hamming distance. Combined with the determination threshold, the DEM data was authenticated. The scope of tampering would be located on the "grid unit" mentioned above. The algorithm has strong robustness against DEM data format conversion, watermarking embedding and other attacks. It is sensitive to various operations of changing contents, and can recognize and locate minor tampering of DEM data. Compared with the traditional DEM authentication algorithm, this algorithm innovatively regards "content" as the sole criterion of identity determination, which effectively compensates for the traditional digital watermarking method’s excessive dependence on information carriers.