# Compute universal threshold threshold = sigma * np.sqrt(2 * np.log(denoised.size))
Parameters: - image: numpy array (grayscale or color) normalized to [0,1] or [0,255] - sigma: estimated noise standard deviation (used for wavelet threshold) - h: non-local means filter strength (larger = stronger denoising) - wavelet: wavelet type for thresholding max denoise
import numpy as np import cv2 import pywt from skimage.restoration import denoise_nl_means, denoise_bilateral from skimage.util import random_noise def max_denoise(image, sigma=0.1, h=1.15, wavelet='db8'): """ Apply maximum-strength denoising using a cascade of methods. # Compute universal threshold threshold = sigma * np
# 1. Strong Non-Local Means (preserves edges while smoothing flat regions) denoised = denoise_nl_means(image, h=h, sigma=sigma, fast_mode=True, patch_size=7, patch_distance=11, multichannel=(image.ndim==3)) 1] or [0
# Apply maximal denoising denoised = max_denoise(noisy, sigma=0.2, h=1.5)
# Display fig, axes = plt.subplots(1, 3, figsize=(12, 4)) axes[0].imshow(original, cmap='gray') axes[0].set_title('Original') axes[1].imshow(noisy, cmap='gray') axes[1].set_title('Noisy') axes[2].imshow(denoised, cmap='gray') axes[2].set_title('Max Denoised') for ax in axes: ax.axis('off') plt.tight_layout() plt.show()