Remote Sensing Velocity Estimation With Misregistration Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing geospatial imagery systems face challenges in accurately estimating the velocity of moving objects due to misregistration between images from different sensors, which leads to errors in velocity estimation.
Innovation Solution
The proposed technology utilizes a satellite system with multiple sensor assemblies to capture images in different spectral bands, aligns these images to correct for misregistration, and applies machine learning to identify moving objects, enabling accurate velocity estimation by subtracting background misregistration from raw velocity measurements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple sensor assemblies are used to capture images in different spectral bands, then the ability to identify moving objects is improved, but misregistration between images occurs leading to velocity estimation errors
Solution Approach 1:
The patent applies preliminary registration to align images from different sensor assemblies before velocity estimation. By pre-aligning the images using identified features and calculating transformation parameters, the system eliminates misregistration errors before they affect velocity measurements, thus resolving the contradiction between using multiple sensors and maintaining measurement accuracy
Solution Approach 2:
The system uses feedback by calculating apparent velocity of stationary features, comparing it against expected zero velocity, and using the difference to correct the velocity estimates of moving objects. This feedback mechanism compensates for registration errors and improves velocity estimation accuracy despite using multiple sensor assemblies
2Measurement precision
If image alignment is performed to correct misregistration, then velocity estimation accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent segments the image alignment process into distinct steps: identifying stationary features in multiple images, calculating transformation parameters based on these features, and applying the transformation to align images. This segmentation makes the complex alignment process more manageable and computationally efficient while maintaining accuracy
Solution Approach 2:
The system creates a transformed version of the original images by applying calculated transformation parameters. This copying approach allows the system to work with both aligned and unaligned data, enabling velocity estimation without permanently modifying the original images, thus balancing accuracy with computational efficiency
Data Source
AI summary
A computer-implemented method is provided for estimating velocity of a moving object. The method includes receiving an image, which includes a first strip generated by a first sensor and a second strip generated by a second sensor, and recognizing a moving object in the first strip of the image. The method also includes selecting a plurality of stationary objects that are captured with the moving object in the first strip of the image, while excluding stationary object(s) in the second strip of the image, wherein the plurality of stationary objects are non-moving features. The method further includes combining misregistration values for the plurality of stationary objects in the first strip of the image and calculating a velocity of the moving object, based on a raw velocity of the moving object and the combined misregistration values.