New Authentication Method for Vector Geographic Data Based on Perceptual Hash
Keywords: Perceptual Hash, Discrete Cosine Transform, Content Authentication, Robustness
Abstract. In the Internet era, vector data can conveniently be stored, distributed and disseminated. This make it convenience for people to use them, and also make it easier for people to edit or modify them, which can reduce the credibility of Geo-information. Traditionally, authentication is done by using a hash function to generate a digital signature for data authentication. However, this method is very sensitive for change even one bit and it is appropriate for accurate authentication such as text. For geographic data, it may experience lossy compression, filtering distortion, geometric transformation, noise pollution, etc. during transmission and application, but it is consistent with the original data perception. Therefore, the traditional cryptography authentication method is not applicable to the robust authentication of geographic data.
Among various authentication techniques, perceptual hashing is a promising solution. Perceptual hashing is a type of unidirectional mapping of multimedia data sets to perceptual digest sets, that is, a multimedia digital representation with the same perceptual content is uniquely mapped into a piece of digital digest, and satisfies perceptual robustness and security. Since a perceptual hash value is a compact representation of the original content, it can be used for robust content authentication of vector geographic data. The advantage of perceptual hashing algorithms over traditional cryptographic hashing algorithms is that they can tolerate differences in quality and format. The same content is always mapped to the same hash value. This is very effective for robust authentication of geographic data.
In this work, we focus on vector geographic data content authentication by perceptual hash algorithms. In the research of vector geographic data authentication, there are usually some authentication methods based on statistics, rough representation of images and extraction of mutation points based on wavelet transform. However, these methods are more or less sensitive to geometric transformation, poor anti-aggression, high complexity, and poor robustness. In order to avoid the shortcomings of traditional authentication algorithms, this paper proposes a vector geographic data authentication algorithm combining DCT and perceptual hashing. The algorithm has high robustness and security against attack, translation, rotation and two types of collusion attacks.