An Improved Median Filtering Image Denoising Algorithm
Yuan Li · 2018
In order to remove the impulse noise in the image more effectively, a new noise detection and noise removal algorithm is proposed. According to the characteristic that the impulse noise has a certain range of pixel values, the 5×5 neighborhood of the pixel to be measured is initially divided into noise points and signal points, and then the four noise points are used to detect whether the noise point is a real noise point. If it is a true noise point, the lower threshold is recorded. All the noise points initially determined are sequentially detected, and the average value of all the thresholds recorded is obtained, and used to detect whether the pixel points initially determined as signal points are true signal points. All the real signal points are directly output and only the noise points are processed. Choose the signal point with the smallest distance in the 3×3 neighborhood of the noise point to be measured. If there is no signal point, perform the denoising in the 5×5 neighborhood. If there is a signal point, replace the measured value with the median of these signal points. The pixel value of the pixel, otherwise the pixel value of the pixel to be measured is replaced with the average value of all the pixels except the pixel to be measured in the 5×5 neighborhood. Experimental results show that the new algorithm not only can effectively remove the noise of the image, but also can better protect the edges and details of the image.