Tempora-spatial-probabilistic model based for mapping paddy rice using multi-temporal Landsat images

Change detection monitoring is an important method of remote sensing classification. This paper proposes Temporal-spatial-probabilistic model (TSPM) to improve the accuracy of classification, which the accuracy was reduced by the existing method, when the remote sensing image was contaminated by cloud. The study area is three countries in LiaoNing province using five Landsat 8 images. The result of TSPM classification is that the user´s accuracy is 92.42%, the producer´s accuracy is 85.62% and the overall accuracy is 86.91%. Thus we conclude that our proposed model (TSPM) is an efficient approach for remote sensing classification.