A Novel Geometric And Radiometric Simultaneous Correction Model (GRSCM) For High-precision Remotely Sensed Imagery

The raw remotely sensed imagery usually contains geometric and radiometric distortions.Therefore,the remotely sensed imagery preprocessing is necessary,before the remotely sensed imagery is utilized.Geometric correction is according to the geometric distortion,radiometric correction is according to the radiometric distortion.With the rapid development of science and technology and the efforts study of many scholars,geometric and radiometric correction technology of remotely sensed imagery has made considerable progress.But in the most remotely sensed imagery preprocessing,geometric correction and radiometric correction is separated.Even ENVI and ERDAS are the most famous international remotely sensed software,and they are widely used.They do not provide methods and functions of geometric radiation synchronization correction.Because the relationship between geometric information and radiometric information of remotely sensed imagery is unity of opposites,so the geometric correction and radiometric correction simultaneous method is more scientific.Because of the relative correction is more common in the practical work,this paper combines many existed preprocessing methods of remotely sensed imagery,then presents and designs a geometric and radiometric simultaneous correction model(the abbreviation is GRSCM)for remotely sensed imagery pre-processing.The model is implemented by programming and is used for imagery correction.The error of correction is analyzed,and the accuracy is evaluated.A new method with good precision for remotely sensed imagery preprocessing is provided,and then the main research contents as follows:1)Geometric and radiometric simultaneous correctionThis study considers the geometric information and the radiometric information of remotely sensed imageries are complementary to each other,and the relationship of them is unity of opposites.At the same time,GRSCM refers the methods of imagery matching,and combines geometric information and radiometric information in the imagery correction.After geometric and radiometric initial coefficients are obtained,the initial coefficients are calculated by mathematical model that was strictly derived.Geometric and radiometric simultaneous correction is implemented.2)Optimization of correction model(1)Geometric correction:rational polynomial model is selected in this study.GRSCM is more general and universal.Considering the high order geometric distortion may not exist in the selected imagery,and over parameterization will lead to accuracy decrease.In order to solve this problem,stepwise regression was used to optimize the geometric correction model(2)Radiometric correction:linear relative radiation correction method is selected in this paper.This method simulates the invariant feature between reference imagery and distorted imagery by linear function.In practice,the linear relationship of gray value between the reference imagery and the distorted imagery needs to be verified.In this paper,the statistical method is used to verify the linear relationship.3)Change area detection(gross error elimination)In the relative correction,generally,the reference imagery and the distorted imagery are obtained in different period.Therefore,features change is inevitable.The feature change wills disturbance imagery correction.It means that the experimental data contains gross errors.At the same time,the feature change brings uncertainty to experimental results.This paper eliminates gross errors by the random sample consensus,and improves the accuracy of correction.4)Accuracy test of remotely sensed imagery correctionIn order to verify the effectiveness of the model,experiment selects different data that is used to evaluate the accuracy of correction.The experimental data includes artificial simulated imageries and remotely sensed imageries.In each set of experiments,common correction method,calculates the correction coefficient and the accuracy of correction.The accuracy of correction is used to compare with the accuracy of method that is proposed in this paper.The accuracies of different methods were determined by comparison.The experimental results show that the geometric and radiometric simultaneous relative correction model,which is presented in this paper,has higher accuracy and better results.It is scientific,accurate and adaptive.The geometric and radiometric simultaneous relative correction model,which is presented in this paper,is suitable for high resolution remotely sensed imagery preprocessing.This paper refers the current remotely sensed imagery correction method,analyses the relationship between geometric correction and radiometric correction is unity of opposites scientific and reasonably,explores the new approach of remotely sensed imagery preprocessing,enriches the theory and method of remotely sensed science.At the same time,the results of GRSCM are calculated in the practical remotely sensed imagery preprocessing work,and then improve the current remotely sensed imagery correction method appropriately.