ass日本风韵熟妇pics男人扒开女人屁屁桶到爽|扒开胸露出奶头亲吻视频|邻居少妇的诱惑|人人妻在线播放|日日摸夜夜摸狠狠摸婷婷|制服 丝袜 人妻|激情熟妇中文字幕|看黄色欧美特一级|日本av人妻系列|高潮对白av,丰满岳妇乱熟妇之荡,日本丰满熟妇乱又伦,日韩欧美一区二区三区在线

Due to the factors such as sensor spatial resolution and heterogeneity of surface features, the mixed-pixels were commonly found in medium-spatial resolution remote sensing data, especially in hilly areas, strong topographic relief, diversity, breakage, mixed distribution and scattered layout of the surface features and other factors constituted the difficulties of remote-sensing image classification mapping. In order to improve the classification accuracy for land use in hilly areas and provide data support for land use monitoring, a combined approach of multiple endmember spectral mixture analysis (MESMA) and random forest (RF) was explored. Based on data source of Landsat-8 operational land imager (OLI) sensor data, the fractional abundance of vegetation, impervious surface and soil was firstly extracted through MESMA. Secondly, totally 20 feature variables were figured out and three combined models were constructed on the basis of data image spectrum, texture and fraction variables to carry out random forest classification experiment. Through comparing between the optimal result from the experiment and SVM and MLC classification results, including the same number of variables, the results indicated that MESMA can derive accurate fraction information. The inclusion of fraction information could help to improve the mapping accuracy of all classification methods (RF, SVM and MLC), which can be up to 90.50%, 88.85% and 86.35%, respectively, the gain of RF classification accuracy was most significant. Comparing with LSMA, the fraction variable generated by MESMA was more useful for improving the accuracy. The combined method of MESMA and RF can achieve the comparatively accurate classification map in the multi-feature variables. The accuracy was better than those of SVM and MLC classification results with the same feature variables. Therefore, the proposed method can obtain high precision in land use classification in hilly area. Based on this method, remote sensing image interpretation of large scales can provide technical support and rational reference for land reclamation monitoring.