Road image filtering algorithm based on improved pseudo-median filter
基于改进伪中值滤波器的道路图像滤波算法*

To address the shortcomings of the existing cellular neural network (CNN)-based filters, such as: slow convergence, poor stability, and relatively vague filtered images, an improved pseudo-median filter was designed, based on difference-controlled CNN. Four improved measures, including changing the value space, introducing random perturbation, expanding the median filter window, and using the Mask method, are proposed and tested in this work. The experimental results demonstrated that the new algorithm can better filter out random pulse noise of various intensities, reserve the image details, and has the good real-time.