Efficient Non-Local Averaging Algorithm For Medical Images For Improved Visual Quality

Image can be distorted by various ways including sensor inadequacy, transmission error, different noise factors and motion blurring. For controlling and maintaining the visual quality level of the image to be very high, it is very important to improve the image acquisition, image storage and image transmission, etc. Achieving high Peak Signal to Noise Ratio (PSNR) is essential goal of image restoration. This involves removing noises present in the image. Non-Local Means algorithm combined with Laplacian of Gaussian filter finds better results and produces good PSNR against impulse noise as well as Gaussian noise. Generally the effect of noise can be reduced using smooth filters for better results. Here, Laplacian of Gaussian (LoG) filter is applied for categorizing the edge and noisy pixels. Before that it is mandatory to obtain local smoothing of pixels. Finally the system performance is improved by averaging the non-local parameters. This is applicable to medical images also for removing impulse noise as well as Gaussian noise. The algorithm has been tested with MRI images and CT images efficiently. Better results are obtained in comparison with the previous methods with respect to better visual quality, PSNR and SSIM.