| The fundamental way out for agriculture is modernization.With the development of global positioning system,satellite remote sensing technology,geographic information technology,communication network technology,computer technology,variable processing equipment and decision support system,world agriculture is developing from traditional agriculture to high-tech agriculture,that is,refined agriculture.UAV remote sensing technology is a rapidly developing earth observation technology in recent years.It has the advantages of high spatial resolution,flight under clouds,fast and flexible,and low cost.It effectively compensates for the limitations of traditional satellite remote sensing in data acquisition.Using drones can quickly shoot aerial images,obtain the required spatial data,and process,model,and analyze in real time.In recent years,the scale of my country’s citrus industry has continued to grow,and it has become the fruit with the highest cultivation area and yield in my country.Serving UAV remote sensing technology in citrus planting production activities to accurately and quickly grasp the growth status of fruit trees in citrus planting areas is of great practical significance for predicting citrus yield,managing citrus growth status,and improving citrus yield.Chongqing is one of the main citrus producing areas in my country,and its citrus resources are very rich.This article takes Wushan County in Chongqing as the research area,selects two orchards in the slope land of Dachang Town and the flat land of Futian Town to discuss the extraction and application of fruit tree information,which are the extraction of fruit tree number,crown width,height and other information and their relationship with Correlation analysis between the yield of individual fruit trees.The article first uses the image data collected by multiple flight heights to be processed by the ContextCapture software and selected by the spatial accuracy test as the basic data,and discusses the impact of the aircraft row height on the spatial data accuracy;then uses eCognition Developer and ArcGIS software for multi-scale segmentation And spatial analysis to extract the fruit tree information,analyze the accuracy difference between the extraction of fruit tree information on flat land and slope land;finally,the feasibility of predicting the yield of a single plant through the morphological information of fruit trees is discussed.The main research results of this article are as follows:(1)The accuracy of the orthophoto plane generated by the images collected by the UAV is affected by the flying height and the surface undulation.In the experiment,10 image control points were selected uniformly in each area,the measured points were used as the verification coordinates,and the orthophotos were used as the verification objects.The results showed that the flying height has a certain influence on the spatial accuracy,and the slope area has a greater influence.Due to the high resolution of the UAV image and the large amount of data,in practice,as long as the error is within the allowable range,comprehensive consideration can be made to select the appropriate flying height to collect data quickly and efficiently.(2)Based on UAV remote sensing,it can effectively extract the information of the number,crown and height of citrus fruit trees,and the extraction accuracy can meet the production needs.The UAV image can be processed to generate DOM and DSM at the same time.It contains rich information and can be deeply excavated.The multi-scale segmentation algorithm and neighborhood analysis method were used to extract the number of plants in the plot,and the accuracy was verified with the visually interpreted data as reference data.The overall accuracy of the neighborhood analysis method for plant number extraction was more than 90%.An object-oriented method was used to extract the crown amplitude value and fit it with the field measurement values.The flat field R~2 was above 0.75,and the slope field R~2 was above 0.69.The extracted value of the height information of the fruit tree is fitted with the measured value,the R~2 of the flat land also reaches more than 0.72,and the R~2 of the slope land is about 0.65.Therefore,the use of unmanned aerial vehicle images and corresponding processing methods can satisfy the need to quickly obtain information about citrus trees at a lower cost.(3)Information such as the height and crown of citrus has a good correlation with its yield.Combined with the information of the number of plants,it can be used to estimate the yield of the orchard.By analyzing the relationship between the yield of a single fruit tree and the measured fruit tree information and the extracted fruit tree information,it is concluded that the shape of the fruit tree has a certain effect on the yield of a single fruit tree.The simple correlation coefficient between the crown area and height and the number of fruit trees per plant They are 0.87 and 0.79,respectively.Using fruit tree canopy area and height information modeling,the R2 between the predicted and actual values of individual fruit trees is 0.7889.It can be used in-depth in the orchard’s estimated production,in order to grasp the orchard’s production situation in real time,and make a reasonable picking and sales strategy. |
