Gaussian White Noise Removal Using Mean Filtering

This document demonstrates the implementation of various image filtering techniques in MATLAB to reduce noise in digital images. The techniques include mean filtering for removing Gaussian white noise, median filtering for eliminating other types of noise, and frequency domain low-pass filtering using ideal low-pass filters. These filtering methods can significantly improve image quality by enhancing clarity and facilitating better analysis. For mean filtering implementation, MATLAB's imfilter function or custom convolution operations can be used with averaging kernels (typically 3×3 or 5×5 matrices) to smooth Gaussian noise. The algorithm works by replacing each pixel value with the average of its neighborhood, effectively reducing random noise variations. Median filtering, implemented using medfilt2 function, is particularly effective for salt-and-pepper noise removal by replacing pixel values with the median of neighboring pixels, which preserves edges better than mean filtering. Frequency domain low-pass filtering involves converting the image to Fourier domain using fft2, applying an ideal low-pass filter (circular mask with cutoff frequency) to attenuate high-frequency noise components, and converting back using ifft2. This method requires careful selection of cutoff frequency to balance noise removal and detail preservation. Selecting appropriate filters is crucial in image processing workflows, as each method has specific strengths depending on noise characteristics and image requirements. Proper parameter tuning ensures optimal balance between noise reduction and feature preservation.