Raw Sensor Image Georegistration for Distortion Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image registration systems suffer from distortions due to uncertainties in sensor location and navigation, limited reference elevation data, and errors in the image registration process, leading to inaccurate image generation and integration of sensor data.
Innovation Solution
A system and method that retains raw sensor data and reprocesses it using improved sensor location estimates and reference imagery, reducing distortions such as foreshortening and layover through georegistration, enhancing image accuracy by iteratively improving inertial state estimates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If image registration is performed using initial sensor location estimates, then image processing can proceed, but distortion and reduced accuracy occur due to uncertainties in sensor location and registration errors
Solution Approach 1:
The patent implements feedback by using the processed image to generate an updated sensor location estimate through georegistration, then reprocessing the raw sensor data with this improved estimate. This closed-loop approach continuously refines the location accuracy, reducing distortion and improving measurement precision in the final image.
2Productivity
If raw sensor data is processed immediately with initial estimates, then processing time is reduced, but image quality suffers from distortions like foreshortening and layover
Solution Approach 1:
The patent performs preliminary processing to generate an initial image and sensor location estimate, then uses this preliminary result to improve the location estimate through georegistration. This preliminary action enables a two-stage approach where quick initial processing is followed by refinement, balancing speed and quality.
Solution Approach 2:
The system uses feedback from the preliminary image processing to refine the sensor location estimate, then reapplies this improved estimate to the raw data. This iterative feedback loop progressively reduces distortions while maintaining efficient processing through staged refinement.
3Device complexity
If limited reference elevation data is used for processing, then processing complexity is reduced, but registration errors increase
Solution Approach 1:
The patent uses feedback from the image registration process to iteratively improve the sensor location estimate even when reference elevation data is limited. This feedback mechanism compensates for the limitations of sparse reference data by continuously refining the registration through multiple processing passes.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method and apparatus for generating an image from raw sensor data. In one embodiment, the method comprises reading the raw sensor data from a sensor, the raw sensor data taken from a sensor at a sensor inertial state, generating an estimate of the sensor inertial state, retaining the raw sensor data while performing steps comprising: generating an image, the image generated at least in part from the estimated sensor inertial state and the raw sensor data and generating an updated estimate of the sensor inertial state, the updated estimate of the sensor inertial state generated at least in part from the generated image and the estimated sensor inertial state. Finally, an enhanced image is generated from the retained raw sensor data and the updated estimate of the sensor inertial state.