Comparative Analysis of Noise Reduction Filters for Medical Images
Lakshmi Kumari, Neetu Mittal, Megha Modi · 2024
For image acquittions and transmission various noises may be introduced, this degrades the visionary properties of images. The noise removal is a crucial and Fundamental step in image processing to enhance the image quality. Gaussian, Salt & Pepper, and Speckle noise are few important type of noise that are most frequently found in medical images. The noise from the image is typically removed during image processing by using filters. Among several filters, the selection can be done on the basis of type of data and the filter's behaviour. The filtering approach will, however, replace this modified pixel value with the average of the pixels nearby, since noise is defined as a sudden change in the pixel value in an image. The two types of filters are spatial domain and frequency domain. The spatial domain filters, also referred to as linear filters, operate directly at the image's pixel level. The frequency of the image is the basis for the nonlinear frequency domain filters. Spatial filters include translating the input image back to the spatial domain after multiplying it by the filter function h(i,j). Frequency filters, on the other hand, carry out the reverse function. For instance, a low pass filter produces a noise-free image by outputting 1 for frequencies below the cut-off frequency and 0 for all other frequencies. These filters can greatly enhance the look and quality of images. In this paper, gaussian, average, median, and weiner filters have been used to remove the noise from the images. The results from different filtering techniques have been presented and compared. The gaussian filter provides the best filtering for noise removal and thus enhances the medical images. Further these images may be useful for doctors and medical practicioners to analyse the disease and diagnose in more effective and efficient manner.