Research On The Geometric Correction Of Remote Sensing Images And The Distribution Law Of Distortion Error

High-resolution remote sensing technology is an important means of detecting the natural environment.It has been rapidly developed because of its rapid,macro,comprehensive,and real-time dynamic characteristics.However,during the process of remote sensing image imaging,it is affected by different factors of the sensor itself and external conditions 5 Leading to geometric deformation of remote sensing images.It is necessary to establish the correspondence relationship(coordinate transformation formula)between the image side coordinates and the geographic coordinates to complete the image geometric correction.Common correction schemes include traditional physical correction models and rational number correction models and their optimization and improvement models.The more mature correction model is the collinear equation correction model.Generally,a rational polynomial correction model is used,and a precise ground control point is used to establish a correspondence between ground features and image graphics to complete geometric correction.In this paper,by acquiring ultra-high-resolution remote sensing data of Xi'an World Park and its north business district,and selecting and measuring 48 ground control points in the experimental area,comparing various correction models and then selecting rational polynomial correction models to compete the experiments.Using the correspondence between the points with the same name in ArcGIS,a large number of geometric correction experiments were performed on the images in the experimental area in groups,and the accuracy and stability of the correction results under different control point layout schemes were statistically analyzed.Through a lot of experimental analysis:(1)Use the rational polynomial correction model to explore the influence of the density of the control points on the geometric correction accuracy of the VHR image with a pixel size of 0.3m.At the same time,the geometric correcti on of the rational function polynomial at different orders is compared.Geometric density correction and accuracy test of different density polynomial correction points under the same order polynomial and different density polynomial correction models under the same density control point,it is concluded that 15 control points are used to perform geometric correction on the VHR image of about 7 square kilometers in the World Park area leads results are ideal.(2)Analyze the accuracy of the correction results by integrating the RMS of the distance errors of all the points with the same name,the distribution of the overrun area and the maximum distortion error value in the correction area,and analyzing the correction results by changing the position of the control point in the same area to explore the stability of these schemes,explored the influence of different control point layouts on the correction results,and then add the economic applicability of control point layouts as constraints,and obtained a correction effect analysis table for different layouts.(3)On the basis of the above research,the error distribution map of the image is made using the checkpoint error and the residual,which provides a reference for the user to use the image with different accuracy.Through the polynomial correction model with different orders for the number of control points participating in the correction and the effect of excess observations on the correction effect,the relationship between the residual value and the overall correction accuracy of the image under different control points is explored.When the number of redundant observations of the polynomial model in different orders is at least more than three groups,the correlation between the root mean square value of the residual value and the overall correction accuracy of the image is high;and it is proposed to use reasonable construction of polygons for Kriging interpolation and control Point construction linear inspection elements to obtain the displacement and angle change of the area,and to accurately and consistently express the distortion error of the image.