传统的DFT(discrete Fourier transform)变换域水印算法中,水印直接嵌入变换后的系数,导致矢量空间数据误差较大。针对此类算法的缺点,本文提出了基于DFT的可控误差矢量空间数据盲水印算法。该算法通过放大处理DFT变换系数,将水印嵌入幅度系数和相位系数中,水印提取采用投票原则,无需原始数据的参与。试验表明,水印嵌入矢量空间数据引起的误差很小;水印对数据格式转换及常见的几何变换攻击具有很好的鲁棒性。
In DFT watermarking algorithms,the watermarksare often embedded directly in the transformation coefficients,causing large errors in watermarked data.To overcome such disadvantage of these algorithms,a blind watermarking algorithm for copyright protection of vector geospatial data based on error reduction is proposed.This algorithm amplifies the DFT transform coefficients and embeds the watermarks in the amplitude coefficients and phase coefficients by DFT transform,and the watermark is embedded several times.In addition,the voting principle is utilized to detect the watermark.No original data is needed in the watermark extraction procedure.The experiments have shown that spatial accuracy of the watermarked data is acceptable and the algorithm is robust to a variety of geometric transformation attacks.