Research Scholar, Anna University, Chennai, Tamil Nadu, India.
Adv Exp Med Biol. 2011;696:471-80. doi: 10.1007/978-1-4419-7046-6_47.
This chapter presents a curvelet-based approach for the denoising of magnetic resonance (MR) and computed tomography (CT) images. Curvelet transform is a new multiscale representation suited for objects which are smooth away from discontinuities across curves, which was developed by Candies and Donoho (Proceedings of Curves and Surfaces IV, France:105-121, 1999). We apply these digital transforms to the denoising of some standard MR and CT images embedded in white noise, random noise, and poisson noise. In the tests reported here, simple thresholding of the curvelet coefficients is very competitive with "state-of-the-art" techniques based on wavelet transform methods. Moreover, the curvelet reconstructions exhibit higher perceptual quality than wavelet-based reconstructions, offering visually sharper images and, in particular, higher quality recovery of edges and of faint linear and curvilinear features. Since medical images have several objects and curved shapes, it is expected that curvelet transform would be better in their denoising. The simulation results show the outperforms than wavelet transform in the denoising of both MR and CT images from both visual quality and the peak signal-to-noise (PSNR) ratio points of view.
这一章提出了一种基于曲波的磁共振（MR）和计算机断层扫描（CT）图像去噪方法。曲波变换是一种新的多尺度表示方法，适用于在曲线处平滑的非连续物体，由 Candies 和 Donoho（Proceedings of Curves and Surfaces IV，France：105-121，1999）提出。我们将这些数字变换应用于嵌入白噪声、随机噪声和泊松噪声的一些标准 MR 和 CT 图像的去噪。在报告的测试中，曲波系数的简单阈值处理与基于小波变换方法的“最先进”技术非常有竞争力。此外，曲波重建比基于小波的重建具有更高的感知质量，提供更清晰的图像，特别是边缘和微弱线性和曲线特征的高质量恢复。由于医学图像有多个物体和弯曲形状，因此预计曲波变换在去噪方面会更好。从视觉质量和峰值信噪比（PSNR）的角度来看，模拟结果表明，在 MR 和 CT 图像的去噪方面，曲波变换优于小波变换。
