Multi-source Data Combination Analyzing Based Crop Discrimination Method Using RS

It is essential to discriminate crops in monitoring crop condition using RS(Remote Sensing). A background database for monitoring crop condition is designed and set up through analyzing operating methods of monitoring crop condition from China and other countries. Combination analyzing method of multi-temporal images and multi-source data is studied in this paper. ,A operating method system for crop discrimination is developed using this method to recognize the crops in Huang huai Hai region as a case study, it can improve the result of crop discrimination, advancing work efficiency, which is used to crop condition monitoring system of Ministry of Agriculture.The background database containing RS,GIS,GPS,statistical and ground truth data is classified to spatial database and attribute database. Management function of the background database is important to other functions such as data gathering, data searching etc. The background database offers powerful working platform and reliable data collected rapidly. The regulations enacted in this study for designing the spatial and attribute database are fundamental to found the background database, which are useful for other application in designing database.A series methods of data combination analyzing are selected to form the operating method system for crop discrimination. Combining GIS, GPS, and other data from field work with RS data can determine interpretation features and set off working regions, combining RS data can enhance spatial features in order to do unsupervised classification efficiently, union of GIS data enable us to join maps and extract features, to analyze crop structure ,crop calendar, cultivating system.The methods and rules in field working first normanized by this study for crop discrimination using RS enable field working to run economically.The small features in crop fields are related to the precision level of crop discrimination using RS. Eliminating the discrimination errors from small features is crucial in estimating area results using RS. A double sampling method is first suggested in this paper in order to solve the problem. The first step is to do a small features sampling in samples using GPS measurements and the total crop area is estimated using the more accurate samples.