Research And Application Of Three Dimention Terrain Model Watermarking In Emergency Rescue System

The rapid development of computer technology and the Internet not only brings convenience to the access and use of digital products, but also rases copyright issues, such as unauthorized usage and modification, copying randomly. Just as images, text, audio and audio, there are issues such as copyright protection, safety certification in Three-dimension terrain model, also known as digital elevation models, which is the important data basis for GIS, entertainment games. In this dissertation, starting from an emergency rescue system based on cloud platforms, some digital terrain models watermarking techniques are studied. The major work and conclusions are listed:Requirement analysis is firstly made in emergency rescue system based on cloud platform, and there are some basic requirements of digital watermark which is used in this system. The design, including topography sampling module, terrain modeling module, and information management module, is made in detail, highlighting the use of digital watermarking in the System.Affine invariance is an important feature of elevation changes rate, and interpolation is widely used in terrain analysis, based on which, a multi-watermark algorithm is proposed:The first watermarking algorithm is to embed watermarking in wavelets base on the analysis of land’s slope. Model is subdivided based on interpolation. Through revising the elevation of interpolation point, the second watermarking is embedded in two-level wavelets of the elevation changes rate of the original sampling point. Experimental result shows that the terrain model embedding this multi-watermarking can keep nearly lossless, and this algorithm is robust against noising, translation, mesh smoothing, and some other attacks. Through the study of the spatial domain image watermark based on SVM, the applicability of SVM on terrain interpolation and image model is analyzed contemporarily, based on which, a multi-watermark algorithm is proposed:SVM can learn the relationship between the interpolation point and its neighboring elevations. Watermarking is embedded by revising the elevation of interpolation point. According to the predication performance of the trained SVM model, the watermarking embedding intensity is determined. Experimental result shows that this algorithm can detect any period change, such as interchange, shear localization or period noise, and locate the tampered area. Meanwhile this algorithm is robust against global noising, translation, mesh smoothing, and some other attacks.