在小波变换的基础上提出了Bandelet变换，可有效地处理高维直线或超平面奇异性问题是该变换的一个特点。基本思路为：采取Bandelet变换对原始图像和水印图像分别进行处理，然后排序迭加相应的Bandelet系数并嵌入水印信息，最后用Bandelet逆变换处理嵌入水印后的图像，继而得到嵌入水印后的图像。相关仿真实验证明，该算法对JPEG压缩、噪声、平滑滤波、剪切等图像失真处理方面具有较强的鲁棒性。

Bandelet transform was put forward based on wavelet transform, which can effectively process high-dimensional line or a hyperplane singularity. The basic idea is that the original image and watermark image are Bandelet transform firstly, then the corresponding Bandelet factor for watermark embedding is sorted and added, and last the watermarked image is processed with Bandelet inverse transform to get the watermarked image. Simulation results showed that the algorithm has good robustness to noise, JPEG compression, shear, smooth filtering and other processing of image distortionh.