Medical image denoising using KPCA with local pixel grouping

Medical images play an important role in diagnosing and treatment of several diseases. For taking better decision and diagnosis of the problem the medical images should be free from noise. Generally noise is added while capturing the image and unexpected movement of subject. Thus image denoising is an important pre-processing step in medical image analysis. The main focus in this paper is performance evaluation of image denoising method such as LPG-KPCA algorithm in medical images. To maintain the local structure quality of an image as much as better, a pixel and its surrounding pixels are treated as a vector variable for getting local similarity features. Here the training samples of vector variable are gathered from the local sliding window in the image. The block matching based LPG-KPCA procedure is iterated one more time for suppressing the leftover noise and for improving the image local structure quality. Experiments on medical images demonstrate the denoising results, used for estimating the performance of the proposed method.