One Medical Image Denoising Method Based On Cartoon-texture Decomposition

Medical image denoising is important in image restoration. With rapid development of medical imaging and high request of health, medical image processing becomes a hot area at home and aboard. Various noises are introduced to make a bad impact on the quality of medical image. They would blur the edges and make the error large and lower SNR to make some disease impossible to be clarified. So protecting the feature of the medical image is of great value while denoising.In this paper, we find one new method to denoise medical image. It uses image cartoon-texture decomposition, image enhancing theory and non-local means method to construct one mixed denoising method. It has more advantages than traditional methods.We first decompose the noising image to two parts:f=u+v. u is the main structure part and v is the oscillating parts which include texture and noise. Secondly, we enhance u to get ul using sharpening method and denoise v to get vl using non-local means filter. Last, we add u1 and v1 to get the new image fl.The numerical experiment expresses that the new method has a higher SNR than the traditional methods and has a better denoising effect. The new method can protect more edges and texture information which can help identify disease. When processing v, we use one new fast non-local means method. This technique helps accelerate the computing speed to 6 times. The new method has higher application prospect.