Impulse noise threshold detection and Noise removal using Non-local Grid Network

The method of removing the existing Impulse noise was to find threshold and apply median filter remove noise. In this paper, we try reference values in neural networks. order noise, used a non-local consider not only information region but also entire region. We then utilized Yolo's grid cell structure make stronger predictions considering various sizes different areas. Accordingly, qualitative quantitative evaluation results show better performance than conventional filters.