| Remote sensing monitoring of crops is always an important topic in the field of remote sensing application.Remote sensing has become one of the most important methods to obtain information in precision agriculture because of its short detection period,large coverage,strong current and low cost.However,due to the characteristics of satellite imaging sensor itself is instantaneous and periodic,and the influence of meteorological conditions on the ground sensor imaging,making it difficult to obtain both high spatial resolution and high temporal resolution data at low cost,remote sensing data space and time granularity restricts it is an important factor in precision agricultural application.Time resolution of spatial resolution and high temporal resolution characteristics of spatial and temporal remote sensing data fusion algorithm can fusion of high spatial resolution remote sensing data,remote sensing data to generate the high temporal and high spatial resolution,has important significance to the application of remote sensing data.At present,most of the spatial and temporal fusion algorithms are based on the fusion of two kinds of remote sensing data.In view of this situation,this paper based on STARFM is put forward using multi-source remote sensing data spatial autocorrelation index improved local weighting function fusion algorithm to make full use of multi-source remote sensing data fusion to improve the quality of data.Moreover,the two methods based on the local autocorrelation index do not need to perform the complex spectral normalization process.The main contents and conclusions of this paper are as follows:1)On the basis of the comparative analysis of the spatial temporal fusion algorithm,based on the STARFM algorithm,combined with the local spatial autocorrelation index,this paper proposes a spatial-temporal fusion algorithm of multi-source remote sensing data.With a high spatial resolution and high spatial resolution data as the main data,the high spatial resolution data for auxiliary data,using high spatial resolution data from the forecast date of last extracted local spatial autocorrelation index improved weight function fusion,with similar texture structure and pixel fusion has more weight,improve the quality of financial data.2)A comparative study of the original STARFM,local Moran’s Index improved STARFM,Getis-Ord local G improved STARFM and based on spectral normalized ESTARFM,with the visual interpretation method and compared using the method of evaluation of quantitative evaluation and analysis of 4 kinds of algorithm fusion accuracy.It shows that the results of local Moran’s Index improved STARFM and Getis-Ord local G improved STARFM are not as good as the results of ESTARFM on quality,but compared to the original STARFM increased significantly.The two methods based on local autocorrelation index do not need to carry on the complex spectrum normalization process,and the application prospect is wider.3)The spectral normalized ESTARFM algorithm first by the target date before and after the two prediction data of high spatial resolution data based on the target date,then the final fusion results obtained by weight combination,the combination weights are fixed for the whole image,inevitably introduces the mixed pixel problem of low spatial resolution.This paper presents a prediction algorithm using STARFM high spatial resolution data of the target date,the time distance weighted average method to predict high spatial resolution data of the target date,then the difference of reflectance target weight combination date of high spatial resolution data with high spatial resolution data before and after the date and the preliminary prediction of each pixel is calculated then,each pixel high spatial resolution data to calculate the target date.4)Because the window size and the estimated parameters of land cover types such as the number of the more obvious to influence the quality of data fusion,the optimal combination of test and analysis of the two parameter fusion of multi-source remote sensing data based on temporal and spatial autocorrelation index method in local space.With the increase of the window,more adjacent pixel information to participate in the fusion process,it will improve the quality of data fusion in a certain range,and when the window increases to a certain extent,there are many with the center pixel independent pixel selected pixel,but will not increase or reduce quality of data fusion,data fusion showed a decreasing trend after the first raising.With the increase of the number of the estimated surface cover types,the pixels in the window are reduced,and the quality of the fused data shows a trend of increasing and then decreasing.In the study area,41 pixels of the window size are selected,and the number of land cover types is estimated as the optimal combination of 2.When the parameter set is greater than or less than the optimal combination,the quality of the fused data is not as good as that of the optimal parameters.5)Heilongjiang Shuangshan farm and the surrounding area as the study area,using multi-source remote sensing data to carry out monitoring and spatio-temporal snow melt crop classification fusion algorithm of the data obtained,and the integration of monitoring and crop classification compared with not using data fusion of snow.The results show that the accuracy of crop classification can be improved and the monitoring ability of snow melting period can be improved by using the data obtained from the multi temporal remote sensing data fusion algorithm. |
PDF Full Text Request