MAE-based Optimized Filter Combining the Median and Multi-scale Filtering Algorithms

As the road image has different types and intensities of noise, filtering the road image noise is essential for robot visual navigation. Taking into account the relatively poor versatility and effectiveness of the general filtering algorithm, and taking advantage of the existing median filter and multi scale adaptive fusion filter, a minimum average absolute error (MAE)-based optimized filter combining the median and multi scale filtering algorithms is proposed. It is found that this method can filter out a variety of common noises more effectively for the road images for an outdoor mobile robot, which enhances the versatility and effectiveness of the algorithm.