A new filtering method based on spatial clustering
基于图像空间聚类的滤波技术

Some traditional image filters may lose detailed spatial information while controlling noise. The remote sensing image cluster algorithm was introduced and improved, and then a spatial clustering method was presented. Spatial clustering method was helpful to distinguish and protect the image spatial patterns, and was used to separate louder noise that affected the image visual effect. Based on the separation of noise points set (NPS) and signal points set (SPS), NPS was filtered and SPS was fused with weighted averaging method. The result shows that new filter based on spatial clustering technique can remove the image noise. Furthermore, it can protect unpolluted detailed spatial information as much as possible. Better filtering result is got on the case of multi-noise.