In the field of remote sensing image change detection,change detection accuracy often cannot be guaranteed when remote sensing images are contaminated by salt and pepper,Gaussian,and mixed noises.Although supervised change detection algorithms based on spatial fuzzy C-means clustering can effectively achieve noise-resistant change detection,their manual training cost and time cost are too high to be applied on a large scale in real-time scenarios.In this regard,this paper couples five spatial fuzzy C-mean algorithms with earth mover's distance(EMD),respectively,to implement five unsupervised remote sensing change detection algorithms with better noise-resistant capability,which can guarantee the real-time change detection accuracy under noise contamination.The experiments prove that compared with the recently proposed KPCAMNet and GMCD unsupervised change detection algorithms,the algorithms proposed in this paper can better process remote sensing images contaminated by salt and pepper,Gaussian,and mixed noises and have certain application values.
维普
万方数据