Research On Algorithm Of X-CT Medical Image Denoising

X-Computed Tomography(X-CT) is an important with magnetic resonance imaging, ultrasound and other medical imaging diagnostic technology, and is one of the important sources of doctors to obtain information. Compared with other methods of medical imaging, X-CT medical image has high resolution of tissue density, small damage to human bodies and other advantages, it is very important to the study of pathology and anatomy. But in the process of the X-CT scan and transfer images maybe appear blurred or boundary unclear phenomenon, thus affecting the X-CT medical image readability, the doctor make accurate diagnosis caused some difficult. Therefore, to eliminate image noise of X-CT medical image method has an important clinical significance and application value.Traditional X-CT medical image denoising method with mean filtering, median filtering, Wiener filter, but because the denoising method in X-CT medical image noise removal at the same time fuzzy image in some important details, the treatment effect has been unable to meet the needs of medical diagnosis.It is widely used in the wavelet transform image denoising method, although in the time-frequency domain multiresolution properties, but its lack of direction, the image edge and detail will produce a certain degree of fuzzy, difficult to complete capture of image contour information, not the best image sparse representation, this will tend to increase the clinical misdiagnosis probability.In order to overcome the above method on X-CT medical image noise component is not analyzed thoroughly and other shortcomings, this paper selects can well capture images of the geometric structure, and can effectively realize the image local, multiple directions, multi resolution of a real image representation method--Contourlet transform.While the Contourlet transform is taken under the sampling operation, the lack of translational invariance, may produce the false phenomenon of Gibbs, make the denoised image distortion, the raw Contourlet transform are improved, it proposes a downsampling Contourlet transform, Contourlet transform to overcome the lack of translational invariance defects.According to this transformation, proposed one kind based on the Context model of non sampling Contourlet transform for X-CT medical image denoising algorithm, and are simulated and analyzed, experimental results show that, compared with other method improves PSNR, the better to retain the image details, can effectively improve the quality of X-CT medical image.Secondly, this paper uses a new method for processing image signal blind signal processing is an important branch of Independent Component Analysis(ICA), this paper proposes an improved independent component analysis based on X-CT medical image denoising algorithm.The method is separated from the point of view, that when a X-CT medical image by the variance of the same noise source pollution after two different, in the ICA separation process, get two new hybrid image; while in the ICA after the separation, get the separation matrix multiplication on the original image can be separated from the mixture of the original image and noise image.The experimental simulation results show that, the algorithm can effectively improve the performance of ICA algorithm and can obtain a higher PSNR, particularly in the image is subjected to noise pollution, with the aid of the reference image noise can be well restored from the original image.