Remote Sensing Inversion Evaluation for High-Value Spatial Consistency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for evaluating the accuracy of remote sensing inversion results are inadequate in assessing spatial consistency and stability, particularly in high-value regions, leading to inconsistent distribution of inverted and measured object content, and are resource-intensive.
Innovation Solution
A quantitative accuracy evaluation method and system that utilizes spatial information similarity by acquiring measured and predicted data, conducting comprehensive index evaluation on high-value regions, including accuracy based on quantity, distance, and area, using geostatistical interpolation for visualization and calculating similarity between high-value regions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If model evaluation indexes (R2, RMSE, MAE) are used to evaluate remote sensing inversion accuracy, then the model accuracy can be tested theoretically, but the spatial distribution consistency and stability of high-value regions cannot be evaluated, resulting in poor visualization results
Solution Approach 1:
The patent divides the evaluation process into two distinct segments: (1) model accuracy evaluation using traditional indexes (R2, RMSE, MAE) on discrete samples, and (2) spatial distribution evaluation using geostatistical interpolation to visualize and compare measured and predicted data across the entire study area. This segmentation allows each evaluation aspect to be optimized independently while maintaining overall evaluation reliability.
Solution Approach 2:
The patent introduces geostatistical interpolation as an intermediary method that bridges the gap between discrete model evaluation and continuous spatial assessment. By interpolating both measured and predicted data to create continuous surfaces, the method enables visualization-based evaluation of spatial distribution consistency, thereby mediating between theoretical model accuracy and practical spatial reliability.
2Reliability
If visual check method is used to compare consistency in positions of high-value regions, then the spatial consistency can be evaluated, but huge manpower and material resources are required when applied to larger preset zones
Solution Approach 1:
The patent replaces the manual visual check method (mechanical human labor) with an automated computational approach using geostatistical interpolation and image processing algorithms. The system automatically generates visualized maps of measured and predicted data, calculates similarity metrics, and evaluates spatial consistency without requiring manual inspection, thereby substituting mechanical human effort with automated mechanical-computational systems.
Solution Approach 2:
The patent transforms the evaluation from a qualitative visual assessment to a quantitative parameter-based evaluation. By calculating specific parameters such as correlation coefficients, mean absolute error, and spatial overlap metrics between interpolated measured and predicted data, the method converts subjective visual judgment into objective measurable parameters that can be automatically computed and compared.
3Productivity
If hash algorithm is used to calculate image similarity for evaluating inversion results, then the calculation can be performed on the whole image, but the consistency and stability testing of abnormal regions (high-value regions) cannot be achieved
Solution Approach 1:
The patent applies local quality by differentiating the evaluation approach for different regions: (1) Overall spatial distribution is evaluated using geostatistical interpolation across the entire study area to assess general consistency, and (2) High-value regions are specifically identified and evaluated with enhanced attention to their spatial patterns, stability, and abnormal characteristics. This allows the evaluation to be both comprehensive in coverage and precise in critical areas.
Solution Approach 2:
The patent performs partial evaluation by focusing computational resources on evaluating high-value regions in greater detail while still maintaining overall spatial assessment. By identifying and specifically analyzing abnormal regions with higher computational effort (multiple interpolation methods, similarity metrics, stability tests), the method achieves excessive action in critical areas while keeping overall evaluation efficient.
Data Source
AI summary
The present disclosure provides a quantitative accuracy evaluation method and system for a result of remote sensing inversion. The method includes the following steps: acquiring measured data of a target element content in a preset zone; acquiring, based on acquired remote sensing image reflectivity of the preset zone, predicted data of the target element content in the preset zone by a remote sensing inversion model; and conducting comprehensive index evaluation of abnormal feature similarity between the measured data and the predicted data. The technical solution adopted in the present disclosure can solve the problem with quantitative accuracy evaluation of remote sensing inversion that inverted object content and measured object content are distributed inconsistently in space.