Quantitative remote sensing inversion prediction result evaluation method and system

Abstract

The invention discloses a quantitative remote sensing inversion result precision evaluation method and system. The method comprises the following steps: obtaining target element content measured data of a preset area; obtaining target element content prediction data of the preset area through a remote sensing inversion model according to the obtained remote sensing image reflectivity of the preset area; and carrying out comprehensive index evaluation of abnormal feature similarity on the actually measured data and the predicted data. By adopting the technical scheme provided by the invention, the inconsistency of the remote sensing quantitative inversion precision evaluation in the spatial distribution of the inversion and actual measurement object contents is avoided.

Description

technical field [0001] The invention belongs to the technical field of remote sensing inversion, in particular to a method and system for evaluating the accuracy of remote sensing inversion results based on spatial information similarity. Background technique [0002] With the continuous improvement of modeling methods and the improvement of the accuracy level of remote sensing data, there are more and more studies on the inversion of element content using remote sensing data; at the same time, in order to evaluate the inversion effect, the accuracy evaluation of remote sensing quantitative inversion results is increasingly widely used . For the evaluation of remote sensing quantitative inversion results, the most important thing is its inversion accuracy, stability and consistency, that is, the evaluation of the similarity between the measured value and the predicted value, including the evaluation of the inversion model, the difference between the measured value and the pr...

AI Technical Summary

Problems solved by technology

[0003] In the prior art, there is an index method for evaluating the accuracy of remote sensing quantitative inversion, using R 2 , RMSE, MAE and other evaluation indicators. These indicators are based on discrete samples and are usually used to test the accuracy of the inversion model. However, this method is mainly based on theory to test the model, so false anomalies cannot be eliminated, resulting in better evaluation accuracy. But the actual visualization results are poor

There is also a method for evaluating remote sensing quantitative inversion results. This method directly uses the method of visual inspection to compare the consistency of the location of high-value areas. This method is only suitable for small-area comparisons, and it will consume a lot of manpower for larger preset areas. material resources

In addition, there is another method to calculate the price based on the similarity of pictures. Generally, a method based on the hash algorithm is used to calculate the similarity between pictures. The Consistency and Stability Test of the Abnormal Areas Needed in the Practical Application of Remote Sensing Inversion, That is, the High Value Areas

[0004] To sum up, most of the methods for evaluating the similarity of the accuracy of remote sensing inversion results in the above-mentioned prior art use model evaluation indicators and field verification. The former cannot evaluate the actual inversion effect, and the latter consumes manpower and material resources. Statistical interpolation is a method to visualize the measured data and predicted data and calculate the similarity of pictures. However, the existing similarity calculation method is based on the similarity calculation of the whole image, and it is impossible to selectively calculate high-value areas according to the demand. The accuracy evaluation of remote sensing quantitative inversion has inconsistencies in the spatial distribution of inversion and measured object content, and has subjectivity and limitations

Images