Image Denoising Based On Finite Ridgelet Transform

Image denoising is widely used in some areas, such as image compression, target recognition, etc. The wavelet threshold is an effective denoising method, suitable for images with smooth areas. But, when it is applied to images with linetype structures, wavelet threshold method introduces too much"ring"effect into the denoised image around the edge of the structure, and thus makes the visual effect even worse than the noisy images. The Finite Ridgelet Transform (FRIT) based denoising methods can keep the details of the image while denoising, but at the same time, introduce the so called"wrap around"effect. Another drawback of FRIT is that it is only suitable for prime size images. In respect to these two drawbacks, this paper demonstrates two methods to improve the FRIT.