在利用曲波对图像进行分解的基础上，提出了一种考虑不同尺度分量的曲波系数间是相互关联的系数概率密度函数(PDF)模型，不同尺度的系数选择依赖其父层系数的大小，从而保证在去除噪声的同时，保留图像的纹理边缘等细节信息，然后对图像进行重构-In the use of curvelet decomposition of the image is proposed based on a consideration of different scale components curvelet coefficients are interrelated coefficient probability density function (PDF) model, different scale factor chosen to rely on the size of the parent layer coefficient in order to ensure the removal of noise while preserving image edges and other texture details, and then reconstruct the image