Studies On Geographical Data Copyright Protection And Integrity Verification

2D digital vector maps, the most important data of Geographical Information System (GIS), have been widely used in the fields of navigation, cadastral management, urban planning, etc. Meanwhile, along with the popularization of the applications like location based services (LBS), trajectory planning, prediction and recommendation, the usage of geographical data has become an indispensable part of people's life.The acquisition of geographical data is an extremely consuming process:it should go though filed survey, digitization, vectorization and coordinate normalization, each of which requires large amounts of human and financial resources. Nevertheless, as digital data, the copy and modification of digital geographical data are very easy. The copyright protection and integrity checking of geographical data become two very important research and industrial problems.Digital watermarking is one of the most important technics for the security consideration of digital data. Generally speaking, watermarking means slightly modifying the underlying data based on a set of well-defined operations to force the data to imply certain secret signal (i.e., watermark). During the detection, the embedded signal can be detected from the watermarked data for copyright assertion and integrity checking. Focusing on the copyright protection and integrity checking of geographical data, in this paper, we design three watermarking schemes. Meanwhile, for the consideration of the fidelity, robustness and fragility of the shcemes, a quality assessment method was proposed to enhance the shcemes' performance. To be specific, this paper mainly includes the following four works:1) Due to the fact that the exisiting works cannot resisit geometric attacks and simplification attack, in this paper, we propose a ruboust watermarking scheme based feature distance ratios for vector maps. The scheme first gives an improved Douglas-Peucker algorithm for the feature point selection of geographic objects to guarantee a stable feature point selection under attacks (esspically for scaling). Then, the feature distance ratio (FDR) is calculated for each object as cover data for the watermark embedding. The bit that should be embedded into an object is decided using the higher bits of its FDR so that correct watermark synchronization can always be guaranteed. Due to the watermark synchronization mechanism, every watermark bit can be covered by multiple objects. Meanwhile, due to the stableness of FDR, the scheme proposed in paper can resist all the robust watermarking attacks, including geometric attacks, simplification, cropping and noise addition. Experimental results show that the scheme is the only piece of work that can resist scaling and simplification simultaneously.2) Current geographical data watermarking schemes mainly foucs on protecting digital vector maps. The protection for newly prevalent data forms like trajectory stream has rarely been concerned. Noticing the fact that trajectory streams are the support data for many important applications like trajectory planning, predication and recommendation, in this paper, we propose a robust watermarking scheme for such data to protect its copyright. In specific, we analysis and define the common modification types (i.e., attacks, including geometric attacks, trajectory compression and segmentation and noise addition) for trajectory streams. Next, we design a single-scan watermarking scheme that first select feature locations from a trajectory stream using time interpolated feature location selection alogrithms, and then embed watermarks into feature distances calculated using the selected feature locations. We design a set of comprehensive experiments based on real world data set. Experimental results show that our shceme can resist the mentioned attacks.3) The exisiting vector map fraile watermarking schemes authenticate data integrity based on vertex/object groups. The authentication granularity is relatively rough, and the authentication result cannot signify which kind of modification type has been applied to the modified group. Consequently, in this paper, we propose a fragile watermarking shceme for vector maps that can detect, localize and characterize data modifications. Before the scheme is designed, the modification types, i.e., object insertion, deletion and update, for objects are defined based on object's geometric structure using turning function. Then, objects are partitioned into various groups based on the quantitive descriptions of their turning functions. Futher, two watermarks, object watermark and group watermark, are embedded into every object. During the detection, all the object and group wateramrks are detected and detection patterns are formed. Based on detection patterns, all the modification types can be detected, localized and characterzed. At last, the fragility of the scheme is analyzed using probabilistic model and verified using experiments.4) Noticing the fact that only location variation has been quantitively evaluated by the current quality assessment methods, in this paper, we propose a quality assessment method from three prospectives, precision of location, geometrical shape of-objects and topological relationships, to quantitively assess a modified map with respect to the original map, in the form of similarity measurement. Then, we use location similarity to control the embedding power so that the location fidelity can be guaranteed. Futher, we use shape similarity to guide the feature point selection to help the robust scheme resist simplification attack. At last, we employ relationship similarity to guide the improvement of robust and fragile watermarking scheme so that the complete fidelity of topological relationship can be achieved. Experimental results show the effectiveness of the method on enhancing the schemes'fidelity, robustness and fragility.