基于小波分形配比的医学图像插值算法研究

Medical image zooming base on wavelet-fractal interpolation

医学图像一般要求边缘细节清晰,而传统的插值方法会导致图像细节信息的丢失.为了满足医学插值放大图像质量和速度的要求,利用小波变换和几何分形的思想,提出了一种改进的几何分形算法和小波变换相结合的图像插值新方法.实验仿真表明,该算法能极大消除细节伪影,同时还保持很好的纹理细节,峰值信噪也得到提高.

The details of medical image must be clear and distinguishable. The quality of interpolated image which zoomed by the traditional method is not desirable. A new interpolation algorithm based on the Wavelet-Fractal Interpolation （WFI） is proposed with more efficient interpolation strategy. The experiment of the medical image suggest that the new algorithm can obtain texture and edge of the image and higher peak signal-to-noise ratio（PSNR） compared with the traditional methods. As a result, the medicol image zooming by the new algorithm is accurate and real.