Zahra Zand - Department of Electrical Engineering, Faculty of Engineering, Razi University, Kermanshah, Iran
Mohammad Reza Ghahri - Department of Electronik Engineering, Malayer Branch, Islamic Azad University, Malayer, Iran
Gholamreza Karimi - Department of Electrical Engineering, Faculty of Engineering, Razi University, Kermanshah, Iran
Meysam Jamshidi - Department of Electronik Engineering, Malayer Branch, Islamic Azad University, Malayer, Iran

in this paper, a bipolar impact noise removal method for color and black-and-white images is presented. In this method, first the noise pixels are identified and then based on a special algorithm, the color code value of the noise pixel is replaced with the value that is calculated based on the pixels of its neighbors. First, the value of each pixel is compared to the numbers zero or ۲۵۵. If the pixel in question had values between zero and ۲۵۵, that pixel is noise-free and no operation is performed on it. If the pixel in question has a value of zero or ۲۵۵, then a ۳ * ۳ window around it is considered and the amount of adjacent pixels is examined. If more than half of the neighboring pixels are not noisy, then the value of the pixel in question is replaced by the median value of its neighbors. If the noise intensity is high, and the number of noise pixels around the pixel in question is more than half the number of neighboring pixels, the window will be larger and the algorithm will eliminate noise and replace the amount of noise pixels in a wider range. Finally, we simulate and evaluate the proposed method in MATLAB software by writing the program code to show its performance. In addition, the proposed method does not show a reduction in performance in noise removal for color images. It should be noted that most of the proposed algorithms for black-and-white images do not work well for color images. The algorithm has a low computational load and is capable of real-time execution.

Pepper-salt noise, Gaussian noise, Middle filter, RGB images, Alpha trim filter, Noise density, Generalized nonlinear filter