Study On The Soil Moisture Predictive Model Based On The Image Shot By Unmanned Aerial Vehicle

Soil moisture is one of essential Parameters in the study of climatology,hydrology,ecology and agriculture,which dominates the transportation and balance of water and heat between lands and air directly. The change of the soil moisture will lead to change of the thermotics characteristic of soil and optical properties of surface,even to the change of climate. The information of the change of land soil moisture in regional and large-scale is important to the research as follows: balance of land-atmosphere interaction,land hydrology,improving the forecast accuracy of regional and GCM,monitoring of flooding and drought,assessment of the growth condition of corps,the natural and ecology Problems. Therefore the research on change and estimation of the soil moisture in regional and large-scale is very important,which also is a key international Problem.The traditional surface observing network can not meet the study on soil moisture dynamic and successive variation in temporal and spatial scale. Furthermore, the widely used microwave measurement though have good penetration, there also have some existence disadvantages, such as the equipment is too heavy, Short-wave range of electromagnetic wave signal is greatly interfered by the atmosphere, and the wave band is difficult to control, which limit their automation in agriculture development. At present, there are a lot of ways to get the soil moisture information by using visual spectrum, near infrared, thermal infrared and other optical remote sensing measures. Although it has small size, easy imaging, short cycle, low cost and in favor of agriculture's future popularization by using of optical remote sensing, this means mostly rely on high-altitude aircraft applications at the present time. Because the band of optical remote sensing can not penetrate cloud, so it has been limited in practical applications. The emergence of unmanned aerial vehicle resolved these issues.As a new way to obtain information, unmanned aerial vehicle along with technology maturity, has been applied in a growing number of fields. Compared with normally used information capture aircraft and satellites, unmanned aerial vehicle is very fitful to the future of agriculture popularization with the chrematistics of low-cost, high accuracy, run a short cycle, easy to operate, etc., This study is based on this, using UAV as aircraft, visual spectrum and near infrared as a means of remote sensing measurements to study how to detect a certain region of the surface soil moisture information ,using Utah, United States for example. In this study, the main contents are as follows:First of all, set up the image of many spatial analysis models and fractional order normalized difference vegetation index (NDVI). visual spectrum as well as near infrared and other information, collected back by unmanned aerial vehicle, are independent. The information associated with soil moisture is insufficient to achieve the prediction standard. Previous studies are analyzed through a variety of formulas, such as HIS, GRAY, NDVI, etc., which integrate this detected information so as to an accurate prediction standard of soil moisture. The purpose of this study by comparing the image characteristics of a variety of spatial analysis, is putting forward a fractional order normalized difference vegetation index, which can greatly improve contains information's correlation coefficients between soil moisture and image, and guarantee the prediction accuracy by using the soil moisture model.Secondly, propose image mosaic algorithm based on local gray-scale matching. There have been a lot of Mosaics Algorithms, which can complete a very good image mosaic mission. However, these algorithms need more time to run, which can not amount to real time requiring by Unmanned Aerial Vehicle (UAV) image mosaic. For this purpose, this paper proposed image mosaic algorithm based on local gray-scale matching. In the process of image mosaic, this study integrated UAV flight characteristics, which greatly narrowed the search area of image matching feature points. And grayscale images for rapid conversion of the image features, in this condition, feature points can accurately and quickly match, saving the operating costs and meeting the requirements of UAV image real-time mosaic.Thirdly, complete the Calibration Algorithm for Unmanned Aerial Vehicle (UAV) image projection. Because Unmanned Aerial Vehicle (UAV) is impact by flight courses, wind speed and other uncertainty factors in the process of its flight, the images will have an uncertainty angle with the ground. This study proposed the Calibration Algorithm for Unmanned Aerial Vehicle (UAV) image projection, which can complete image angle correction automatically, based on the information recording by Unmanned Aerial Vehicle (UAV) immediately, and laid a foundation for the image mosaic.Fourthly, propose image segmentation model based on cluster analysis algorithm. It is very important to the soil moisture description as well as information extraction of agricultural automatic irrigation throng determining the region border of image. The usual segmentation algorithms are classified according by the distribution probability of pixels or image features and spatial analysis, which results are rough. In this study, it will be better to zoning boundaries through the cluster analysis algorithm, and according to the gradient of boundary points, as well as its distribution trends.